Computer science papers from the Papers We Love repository. https://paperswelove.org/papers/ Papers We Love 2026-01-24T23:47:00+00:00 https://paperswelove.org/papers/a-calculus-of-communicating-systems-ff4d97c0/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Robin Milner <h3>Abstract</h3><p>This paper introduces a theoretical model for understanding and analyzing systems that involve interacting components. The calculus, known as CCS (Calculus of Communicating Systems), provides a framework for modeling the communication and synchronization between concurrent processes. The model aims to facilitate reasoning about the behavior of composite systems which are constructed from individual components. Significant emphasis is placed on the algebraic laws that capture the behavior of these systems, enabling formal verification and analysis.</p><p><a href="https://moodle.risc.jku.at/pluginfile.php/3407/mod_resource/content/1/A%20Calculus%20of%20Communicating%20Systems%5B1980%5D.pdf">Read the paper</a></p> https://paperswelove.org/papers/fades-fine-grain-adapting-b-trees-by-splitting-and-feb4ec0c/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Feng Chen, Fu Chen, Youjip Won <h3>Abstract</h3><p>Indexing is crucial for efficient data access in database systems. Traditional B-trees, although widely used, are not always adaptable to dynamic workloads. FADES is a novel fine-grained adaptation strategy for B-trees, which employs dynamic splitting and merging techniques to optimize index performance under varying workloads. The proposed method adjusts the node capacity dynamically to maintain a balanced search performance while minimizing the height of the tree. FADES is demonstrated to provide significant performance improvements over conventional B-trees, particularly in fluctuating workloads and rapidly changing data distributions.</p><p><a href="http://home.in.tum.de/~ding/files/fades-pre-camera.pdf">Read the paper</a></p> https://paperswelove.org/papers/preventing-tls-man-in-the-middle-attacks-in-real-t-fe547893/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 J. Alex Halderman, Matthew E. Bernhard, Megha Madhusudhan, Rachel Cummings, Xiaozhu Meng <h3>Abstract</h3><p>Man-in-the-middle (MITM) attacks pose a serious threat to real-time Voice-over-IP and video networks. Internet routers are frequently misconfigured, violating the assumptions of threat models that underly security protocols like TLS. As a result, these security protocols are often incorrectly implemented in domains like video capture, video mixing, party joining, video censorship, video mixing, and other web-based communication functionality. MITM attacks against routers can result in damage ranging from misconfigured routers to full interception and AES or TLS key leakage. We propose a method for detecting TLS MITM attacks over routers, based on incorporating real-time traffic analysis to detect key leakage or attempts to impersonate AES-GCM traffic, TLS ciphertext, and other video content. We precisely describe a resource-efficient means for configuring systems that detect MITM attempts in production communications environments, discussing lightweight mechanisms to detect attacks in real-time, and techniques to verify and alert on misconfigurations before they can be exploited. Our evaluations are performed on various routers, and we demonstrate our approaches in a variety of popular network configurations.</p><p><a href="https://www.usenix.org/node/189029">Read the paper</a></p> https://paperswelove.org/papers/hydra-the-kernel-of-a-multiprocessor-operating-sys-fa21e695/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Harley Schwetman, Roy Levin, Sue McJones, William Wulf <h3>Abstract</h3><p>HYDRA is a kernel for a multiprocessor operating system. Its prime purpose is to provide a setting for the study of a new approach to the design of systems based on capabilities for protection of information and resources. To this end, it provides a facility for expressing and enforcing a wide variety of policies. These policy defining facilities, coupled with the ability to test for changes to the kernel state triggered by requests for services, make it possible to tailor the system for specific purposes. Thus HYDRA is not, of itself, an operating system, but a kernel on which to build operating systems. With such flexibility it also allows for the creation of a wide variety of resource management policies, thereby encouraging experimentation of new approaches to system organization and control. HYDRA, as a tool for such experimentation, provides examples of special purpose kernel-based systems that have been derived from it, showing the adaptability of its architecture. The philosophy and structure of HYDRA are described and an indication of how an operating system might be realized from the capabilities-based kernel is given.</p><p><a href="https://kilthub.cmu.edu/articles/HYDRA_--_the_kernel_of_a_multiprocessor_operating_system/6606224/files/12096734.pdf">Read the paper</a></p> https://paperswelove.org/papers/spectre-attacks-exploiting-speculative-execution-f93adbbe/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Daniel Genkin, Daniel Gruss, Michael Schwarz, Mike Hamburg, Moritz Lipp, Paul Kocher, Stefan Mangard, Thomas Prescher, Werner Haas, Yuval Yarom <h3>Abstract</h3><p>Modern microprocessors use speculative execution to improve performance by predicting program behavior and executing instructions ahead of time. This performance optimization was considered to be invisible to program execution. In this paper, we show that speculative execution can be exploited to leak information across security boundaries, even without accessing privileged memory, such as when code is executed in a sandbox. We demonstrate attacks on (1) Intel, AMD, and ARM processors, and (2) various other inversions of Spectre-based vulnerabilities. We also discuss how to implement software countermeasures to make code resistant to such attacks.</p><p><a href="https://spectreattack.com/spectre.pdf">Read the paper</a></p> https://paperswelove.org/papers/one-vm-to-rule-them-all-f8e77f8b/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Andreas Woergoetter, Andreas Woess, Christian Wimmer, Laurence Tratt, Michael Franz, Michael Haupt <h3>Abstract</h3><p>Virtual machine-based execution of programming languages enables engine implementers to place common functionality into a reusable component. The Java Virtual Machine (JVM) is a popular choice, but was originally designed for statically-typed languages like Java. Many developers have extended the JVM to support dynamically-typed languages like Python and Ruby, but their implementations do not take advantage of the full range of type information available. One result is that interfaces between languages are clumsy and slow compared to pure JVM languages. We propose a new approach to implementing dynamically-typed languages on the JVM, by generating optimized machine code on-the-fly. We evaluate the performance, interoperability, and ease of implementation with existing approaches, and argue that our solutions lead to significant benefit for language implementers and users.</p><p><a href="http://lafo.ssw.uni-linz.ac.at/papers/2013_Onward_OneVMToRuleThemAll.pdf">Read the paper</a></p> https://paperswelove.org/papers/jitsu-just-in-time-summoning-of-unikernels-f79bc70e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Amir Chaudhry, Anil Madhavapeddy, Balraj Singh, Dave Sheets, David J. Scott, Jon Crowcroft, Jon Ludlam, Richard Mortier, Steven Smith, Thomas Gazagnaire, Thomas Leonard <h3>Abstract</h3><p>Unikernels allow us to specialize virtual machines to a far greater extent than traditional hypervisor and container systems. While previous results have shown the benefits of specialization to the memory footprint and boot time of a VM, we show that the effect on both network traffic and latency is even more important. This paper presents a new live cloud orchestration system called Jitsu that uses existing unmodified APIs to reduce network latency by an order of magnitude for short-lived applications and services. We show that, when run in conjunction with our MirageOS unikernel toolchain, Jitsu can uniquely be deployed in environments requiring extreme scale isolation without sacrificing performance.</p><p><a href="https://www.usenix.org/system/files/conference/nsdi15/nsdi15-paper-madhavapeddy.pdf">Read the paper</a></p> https://paperswelove.org/papers/proling-audio-visual-conferencing-sessions-for-qua-f706c272/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 I. Sodagar, J. Adams, K. Ramakrishnan, P. Pancha <h3>Abstract</h3><p>This paper presents a system for profiling and assessing the quality of experience (QoE) for audio-visual conferencing sessions. The system uses a combination of network and media metrics that capture the performance characteristics of the session. By analyzing audio-visual media quality, packet loss, delay variation, and bandwidth utilization, the system provides a comprehensive assessment of QoE. Experimental results demonstrate the efficacy of the proposed system in identifying sessions with poor QoE and the potential for optimization. The study concludes with suggestions for future improvements and enhancements in assessing QoE in conferencing applications.</p><p><a href="https://research.google/pubs/pub48190/">Read the paper</a></p> https://paperswelove.org/papers/a-design-method-for-computer-security-testing-f2bc5f88/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 H. L. Owen, R. D. Tenny <h3>Abstract</h3><p>This paper posits a methodological process for testing computer security, focusing on organized testing methods. It presents both reactive and proactive testing aspects, describing a comprehensive testing strategy to uncover security vulnerabilities. The design method is elucidated with specific considerations for different phases and areas of testing.</p><p><a href="https://core.ac.uk/download/pdf/35074556.pdf">Read the paper</a></p> https://paperswelove.org/papers/robust-physical-world-attacks-on-deep-learning-vis-ee85b158/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Amir Rahmati, Atul Prakash, Bo Li, Chaowei Xiao, Dawn Song, Earlence Fernandes, Ivan Evtimov, Kevin Eykholt, Tadayoshi Kohno <h3>Abstract</h3><p>Deep neural networks (DNNs) have been widely adopted in various visual classification applications. Despite their great success, recent studies found that DNNs are vulnerable to adversarial examples in the digital world, where inputs are artificially perturbed to mislead DNNs into making incorrect predictions. In this work, we demonstrate that physically realizable adversarial perturbations can be generated to attack DNNs in the physical world. We develop robust attack algorithms that are capable of achieving this goal. We empirically validate the effectiveness of our attacks under various conditions, including different physical environments and object distances, and evaluate potential defenses.</p><p><a href="https://dl.acm.org/doi/pdf/10.1145/3267809.3267845">Read the paper</a></p> https://paperswelove.org/papers/hyperbolic-3-manifolds-that-fiber-over-the-circle-edded426/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 C. D. Hodgson, S. A. Bleiler <h3>Abstract</h3><p>This paper studies the geometry of hyperbolic 3-manifolds which fiber over the circle, with the fibers being surfaces with punctures. It is shown that the complete hyperbolic structure on these manifolds can be described by a simple canonical ideal triangulation.</p><p><a href="https://www.maths.ed.ac.uk/~v1ranick/papers/francisweeks.pdf">Read the paper</a></p> https://paperswelove.org/papers/how-do-they-pay-you-for-longer-articles-ebf99671/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Eric Gilbert <h3>Abstract</h3><p>We use Mechanical Turk to measure what people believe online writers should be paid for longer articles. We find that readers (1) generally think an article should be a little over twice as expensive to write when it’s roughly five times as long; (2) tend to believe that someone writing a 25-paragraph piece should earn $90, on average; and (3) expect editors to pass along cost savings associated with shorter articles to readers.</p><p><a href="https://dl.acm.org/citation.cfm?id=2812806">Read the paper</a></p> https://paperswelove.org/papers/the-new-threat-passive-online-attacks-against-pkcs-e7de2f4a/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 D. Armstrong, Rahul Kholia <h3>Abstract</h3><p>Devices using PKCS#11 middleware are extremely widely deployed both in standalone modes and as part of more complex structures such as smart cards and Hardware Security Modules (HSMs). Despite this, previous work on practical attacks against these devices has so far been very limited. In this paper, we examine how side-channel information can be combined with more traditional cryptanalytic skills to mount a passive attack against PKCS#11 devices. This passive attack differs substantially from previous attacks against keystores, both in terms of the underlying assumptions, and in the attack vector utilised. We believe that our attack sits above previously known vulnerabilities, both in terms of novelty and scope.</p><p><a href="https://www.usenix.org/system/files/conference/woot13/woot13-kholia.pdf">Read the paper</a></p> https://paperswelove.org/papers/all-you-need-is-ears-a-multi-sensory-embodied-agen-e6d0fef5/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adrià Colomé, Carme Torras, Felix Gimeno, Marta R. Costa-jussá, María García Ortiz <h3>Abstract</h3><p>We propose a multi-sensory neural network that enables an embodied agent to process aliased audio-visual observations in partially observable environments. Working towards this goal, we introduce a semi-supervised approach to enhance linguistic capabilities by combining audio and visual information, and illustrate its ability to disambiguate language conditioned on visual context. We present an exhaustive set of experiments demonstrating how the fusion of audio and visual sensory signals helps to perform challenging tasks like audio-visual speaker diarization and speech recognition. These experiments reveal that our proposed approach exceeds single-modality models in segments with multi-sensory input, this being particularly relevant for real-world applications where observations are often ambiguous and carry inherent uncertainty.</p><p><a href="https://arxiv.org/pdf/1901.04592.pdf">Read the paper</a></p> https://paperswelove.org/papers/efficient-privacy-preserving-stream-aggregation-in-d54feb63/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Guojun Wang, Junggab Son, Shuai Hao, Weisong Shi <h3>Abstract</h3><p>Mobile sensing relies on users to constantly contribute personal data from sensors embedded on smartphones or wearable devices. Privacy concerns, however, prevent the extensive exploitation of the mobile sensing technology. Traditional cryptographic techniques, such as homomorphic encryption, alleviate privacy concerns but are not feasible for battery-powered, resource-constrained mobile devices due to high computation overhead. In this paper, we propose an efficient privacy-preserving stream aggregation (EPSA) scheme that protects individual user's data privacy while allowing an aggregator to collect and aggregate user-contributed data with low aggregation error. To achieve this, we have designed EPSA based on a novel homomorphic encryption scheme tailored for mobile devices. We analyze the performance and security of the scheme, and demonstrate its efficiency through real evaluations on both smartphones and programmable logic controllers. Our evaluation results show that EPSA effectively reduces the computational cost and energy consumption of mobile devices, while preserving data privacy with low aggregation error.</p><p><a href="http://www.eecs.qmul.ac.uk/~tysong/files/IFIPNetworking14.pdf">Read the paper</a></p> https://paperswelove.org/papers/towards-understanding-the-nature-of-architectural--d54a1c21/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Carnegie Mellon University Software Engineering Institute, M. P. Singh, Robert L. Nord <h3>Abstract</h3><p>This paper reports initial work investigating ways to characterize and understand problems of software architecture. We identify elements of architectural problems, such as requirements and technological constraints, and show how they can be used to define the solution domain. We propose a framework that organizes architectural problems based on notions of clarity and difficulty. We further illustrate the framework with examples, in the hope that it will lead to useful methodologies for architectural problem analysis.</p><p><a href="https://dl.acm.org/citation.cfm?id=122124">Read the paper</a></p> https://paperswelove.org/papers/digital-storytelling-for-reflection-and-engagement-d537333c/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Andrew Morrison, Gry Wennberg, Mads Hobye <h3>Abstract</h3><p>This paper discusses an exploratory study of digital storytelling in work organizational settings. It is based on design projects undertaken with different work groups in Norway. The paper outlines approaches to the research and use of digital storytelling as well as issues relating to developing an understanding of the frames of digital interaction and design. It also considers how digital storytelling might be applied as a means for encouraging self-reflection and enhancing work engagement.</p><p><a href="http://oda.oslomet.no/oda-xmlui/bitstream/handle/10642/3189/1321835.pdf">Read the paper</a></p> https://paperswelove.org/papers/ix-a-protected-dataplane-operating-system-for-high-d449d333/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adam Belay, Christos Kozyrakis, Costin Iancu, David Mazières, Eric Eide, George Prekas, Kai Keller, Marton Berezecki <h3>Abstract</h3><p>High throughput and predictable low latency in a dataplane operating system is achievable by using techniques that separate the control plane from the data plane, access application memory using virtual memory mechanisms, and apply lightweight, cooperative context switching. We built IX, a dataplane operating system designed for high I/O and processing efficiency while maintaining the protection between the control and data planes. IX shows significant performance improvements over traditional operating systems and is made publicly available to ensure reproducibility and further research.</p><p><a href="https://infoscience.epfl.ch/record/223617/files/tocs16-ix.pdf">Read the paper</a></p> https://paperswelove.org/papers/uniprof-online-analyzis-of-application-performance-d12f846e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Florian Schaffert, Stefan Schmid <h3>Abstract</h3><p>Application performance analysis is complicated today because applications run in complex environments. To identify and to fix performance problems, it is vital to be able to understand an application's resource consumption behavior and its interactions with the system in detail. We present UniProf, a novel online performance analysis approach. Contrary to existing work, UniProf does not require specific profiling instrumentation, nor does it build on (offline) tests. Rather, UniProf provides a universal and continuous analysis of the application in its real environment and in real time. UniProf can detect costly resource interactions online, and provides actionable feedback on how to optimize applications. Moreover, the feedback is provided in real-time, simplifying debugging and performance improvement efforts.</p><p><a href="https://flosch.eu/papers/2017-sigcomm-uniprof.pdf">Read the paper</a></p> https://paperswelove.org/papers/diversity-an-objective-measure-of-semantic-distanc-cf29cfd5/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Shailendra Raj Mehta, Sudheendra Kambhampati Kannan <h3>Abstract</h3><p>Semantic Web ontologies offer the potential to provide more precise answers to complex queries than statistical search engines. The labeling of semantic content can support similarity inference, improving searching and information retrieval quality. In this paper, we argue for the development of a measure of diversity that quantifies the semantic distance between concepts in an ontology. We define such a measure based on an extension of traditional methodologies in distance and similarity measurement across taxonomical trees. This measure is intended to help distinguish between semantic properties accurately, and facilitate innovative solutions in diverse areas such as knowledge representation in artificial intelligence, natural language processing, and diverse web applications.</p><p><a href="https://www.usenix.org/events/hotos11/tech/final_files/Kannan.pdf">Read the paper</a></p> https://paperswelove.org/papers/an-introduction-to-capsicum-cee41fc4/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Justin Cormack <h3>Abstract</h3><p>Capsicum is a lightweight operating system capability and sandbox framework developed at Google Research and the University of Cambridge. It targets application compartmentalization: the decomposition of monolithic UNIX applications into logical applications into logical applications into logical application components, providing increased application and system security. Capsicum blends the best aspects of capability systems, which have not previously achieved mainstream adoption, with the best aspects of POSIX (UNIX) APIs, so as to incrementally deployable. In its hybrid capability model, processes may interleave capability mode and ambient authority mode. We provide an introduction to capability systems, before exploring Capsicum’s hybrid capability architecture and application to UNIX applications, focusing on the file descriptor as a capability, and the semantics of capability mode. We describe the design and implementation of two core OS components: capability mode and capabilities, and UNIX process descriptors (capabilities-safe process identifiers). Finally, we explore adapting core FreeBSD utilities to support capabilities and the Capsicumization of ongoing work on UNIX applications, such as Chromium.</p><p><a href="https://www.netbsd.org/gallery/presentations/justin/2015_AsiaBSDCon/justincormack-abc2015.pdf">Read the paper</a></p> https://paperswelove.org/papers/explaining-inconsistencies-in-preferences-using-we-cba23f5e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 David C. Parkes, Felix Fischer, Revital Lavi <h3>Abstract</h3><p>We introduce a model for preferences that is based on weighted goals, motivated by applications to detecting and resolving inconsistencies in preference judgments. In this model, preferences are constructed from a linear order on a set of basic elements, such as items or alternatives, and a set of weighted goals. Preferences induced by a weighted-goal model can lead to inconsistent judgments across pairs of alternatives, and we present an approach to identify and explain such inconsistencies. We also study algorithms for identifying all inconsistent preference judgments and discuss the complexity of the problem, providing both positive and negative complexity results.</p><p><a href="https://arxiv.org/pdf/1904.09225.pdf">Read the paper</a></p> https://paperswelove.org/papers/hypervisors-vs-lightweight-virtualization-a-perfor-c8455d59/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Antonella Molinaro, Francesco Risso, Ivano Cerrato, Mohammed Ilyas, Roberto Morabito, Torsten Braun <h3>Abstract</h3><p>This paper presents a performance comparison between hypervisors and lightweight virtualization technologies, with a specific focus on compute, memory, and network performance. To achieve this goal, different testbeds have been set up, including an OpenStack Cloud-based environment. The outcomes highlight the different overhead levels introduced by the tested virtualization solutions and determine how this impacts the achievable performance.</p><p><a href="https://www.researchgate.net/profile/Roberto_Morabito/publication/273756984_Hypervisors_vs_Lightweight_Virtualization_A_Performance_Comparison/links/550a83660cf26198a63afb10.pdf">Read the paper</a></p> https://paperswelove.org/papers/scalable-loss-recovery-algorithms-for-reliable-mul-c45e100b/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Samuli Turtinen <h3>Abstract</h3><p>The thesis focuses on loss recovery algorithms in reliable multicast protocols. Multicasting is a method of information delivery to multiple recipients at once. It is often applied to minimize network resources when sending the same data to several receivers. However, in many cases, multicast protocols do not guarantee reliable delivery: some packets are lost and not retransmitted. Reliable multicast protocols, offering guarantees similar to point-to-point reliability, have been studied intensively. The main challenge is loss recovery since retransmitting lost packets in a scalable manner remains difficult. This thesis contributes to the field by presenting novel loss recovery algorithms and techniques for use in reliable multicast protocols. The proposed methods aim to improve scalability and efficiency in packet loss recovery in different network environments. The algorithms have been evaluated using both theoretical analysis and simulations. The results show improved performance over existing methods, indicating that the proposed solutions present a viable alternative for efficient and scalable loss recovery in reliable multicast.</p><p><a href="https://aaltodoc.aalto.fi/bitstream/handle/123456789/6318/isbn9789526049175.pdf?sequence=1">Read the paper</a></p> https://paperswelove.org/papers/the-information-content-of-bank-exam-ratings-and-s-c296c986/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Cynthia A. Glassman, Donald P. Morgan, Michael W. James <h3>Abstract</h3><p>This paper investigates the ability of subordinated debt markets to assess banks' conditions and predict bank ratings from regulators. Using a sample of large holding companies over the 1987-1992 period, we find that subordinated debenture yields, adjusted for the level and term structure of interest rates, have information content about bank holding company exam ratings. However, they do not predict these ratings as well as an econometric model using public data. This suggests that while market data provides useful information, it does not appear to predict regulator ratings better than publicly available data.</p><p><a href="https://apps.dtic.mil/dtic/tr/fulltext/u2/a293537.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-multikernel-a-new-os-architecture-for-scalable-bf5a32f7/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adrian Schüpbach, Ales Tumanov, Andrew Baumann, Paul Barham, Pierre Evely, Rebecca Isaacs, Simon Peter, Tim Harris, Timothy Roscoe <h3>Abstract</h3><p>This paper presents the multikernel model for operating system (OS) structure that is designed to make OSes scalable across many cores and software isolated, to enable easier development using message-passing. The key observation is that existing shared-memory approaches inherently lose performance as the number of cores increase over non-shared architectures. The multikernel OS breaks with established shared memory organization to treat the OS as distributed regardless of the hardware. We have implemented a prototype multikernel OS, Barrelfish, to explore the consequences and benefits of our design decisions and to validate our approach through performance experiments.</p><p><a href="http://www.cse.iitd.ac.in/~sbansal/csl862-os/readings/barrelfish.pdf">Read the paper</a></p> https://paperswelove.org/papers/applications-of-topological-methods-in-the-analysi-bd202dca/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Afra Zomorodian, Alberto Carlsson, Anne Collins, Robert Ghrist <h3>Abstract</h3><p>This paper surveys recent work on applications of topological methods for the analysis of sensor networks. Our aim is to show how a variety of topological techniques, both classical and computational, are being developed and used in the areas of coverage, routing, and network resilience. This paper demonstrates some novel applications of elementary algebraic topology, including applications of obstruction theory, to distributed coverage verification, homologies to detect boundaries in sensor networks, and sheaf cohomology for data processing. We also describe how computational tools from topological data analysis, such as persistent homology, can be employed to infer global information from local data.</p><p><a href="https://www.math.upenn.edu/~ghrist/preprints/ATSN.pdf">Read the paper</a></p> https://paperswelove.org/papers/active-networks-and-active-network-management-a-pr-baf999fa/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Ellen Zegura, Ken Calvert, Samrat Bhattacharjee <h3>Abstract</h3><p>Active networks represent a novel approach to network architecture. They support user-injection of customized programs into network nodes, thereby enabling applications to manage and manipulate their own data or control traffic in the network. We define an "active network management" framework to support active networks, based on state management and control mechanisms. By supporting dynamic loadable management applications, active management provides a vehicle that operators can use to automate and control network and service management. We present a case study that highlights some of the issues in building, deploying, and using active management, and argue that it represents an important, viable approach to modern network management issues, particularly in large-scale, distributed networks.</p><p><a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.52.4651&rep=rep1&type=pdf">Read the paper</a></p> https://paperswelove.org/papers/flipping-bits-in-memory-without-accessing-them-an--ba7641d7/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Chris Fallin, Chris Wilkerson, Donghyuk Lee, Jeremie Kim, Ji Hye Lee, Konrad Lai, Onur Mutlu, Ross Daly, Yoongu Kim <h3>Abstract</h3><p>As DRAM cells continue to scale down to smaller dimensions, they become more vulnerable to interference. One such form of interference, referred to as disturbance errors, can be induced in cells that are physically proximate to cells being frequently accessed. In this work, we show that it is possible to induce disturbance errors in most DRAM memory modules (more than 80 percent of the available test samples) of three major DRAM manufacturers by performing reads to only one address. We call this phenomenon Row Hammer. We analyze the properties of these disturbance errors experimentally and provide a comprehensive characterization of the Row Hammer phenomenon. Our findings indicate that these errors are the results of the interaction between susceptible DRAM cells and aggressive memory access patterns. Our study shows that manipulating the physical memory layout and the access pattern might be exploited by malicious software to cause disturbance errors, which can potentially lead to compromise of system integrity in certain systems. This paper provides a brief overview of our experimental methodology, summarizes our experimental results, and discusses solution and future research directions to reduce DRAM disturbance errors induced by Row Hammer.</p><p><a href="https://people.inf.ethz.ch/omutlu/pub/dram-row-hammer_isca14.pdf">Read the paper</a></p> https://paperswelove.org/papers/evaluating-bert-for-natural-language-inference-a-c-ba32f240/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Brian Lester, Daniel Pressel, Kathleen McKeown, Ramesh Nallapati <h3>Abstract</h3><p>In this paper, we evaluate BERT, a language representation model, for solving the Natural Language Inference (NLI) task. We report results on MultiNLI and SNLI datasets, comparing BERT to established methods as well as among the component variants of BERT. Additionally, we performed a more in-depth analysis by investigating the model’s behavior on the vampires and humans sections from Dracula, finding that BERT learns about the semantic properties of the texts, preserving coherence within the narrative.</p><p><a href="https://arxiv.org/ftp/arxiv/papers/1904/1904.08239.pdf">Read the paper</a></p> https://paperswelove.org/papers/teaching-software-engineering-for-game-development-b7a73996/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Caitlin Sadowski, James R. Larus, Kathi Fisler, Shriram Krishnamurthi, Toby Hooper, William F. Marrero <h3>Abstract</h3><p>As game development becomes an accepted academic field, CS departments will face several challenges analogous to those faced in software engineering teaching a decade ago. This paper explores what a course on game development should cover, proposing an experience grounded in software engineering principles. We also discuss our experience in teaching such a course for the first time.</p><p><a href="https://cs.brown.edu/~sk/Publications/Papers/Published/cgkmf-teach-gc/paper.pdf">Read the paper</a></p> https://paperswelove.org/papers/taxonomy-of-optimization-algorithms-b6fdd2c4/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 William M. Farmer, Yvon Gauthier <h3>Abstract</h3><p>The optimization of complex systems is an important problem in many fields including engineering, economics, and computer science. This paper presents a taxonomy of optimization algorithms by categorizing them based on common characteristics. We propose that this taxonomy can be a useful tool for choosing an appropriate algorithm for solving a specific optimization problem.</p><p><a href="http://www.cas.mcmaster.ca/~bill/best/algorithms/07Taxonomy.pdf">Read the paper</a></p> https://paperswelove.org/papers/breach-ssl-gone-in-30-seconds-b5cec472/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Angelo Prado, Nabil Schear, Yoel Gluck <h3>Abstract</h3><p>The data compression ratio of HTTP responses is the basis for our attack. Together with the ratio, we maliciously breach plain-text data based on a vulnerability present in the SSL/TLS protocols. This defect affects all versions of the SSL/TLS protocol that use HTTP as the application layer protocol and compress data prior to encryption. Attackers can recover secret authentication cookies or other secret bearer tokens, effectively as a side-channel attack. Our attack on HTTP compression compromises the forward secrecy and breaks an authenticated SSL connection. In addition, we developed a mitigative technique: separating the compression layer from session state within the HTTP response. We believe that more work needs to be done to fully mitigate the dangers outlined in this paper.</p><p><a href="http://breachattack.com/resources/BREACH%20-%20SSL,%20gone%20in%2030%20seconds.pdf">Read the paper</a></p> https://paperswelove.org/papers/cooperative-sequential-adsorption-ae9ad613/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 David Andelman, David J. Tannor, Haim Diamant, Tal Cohen <h3>Abstract</h3><p>We study one-dimensional cooperative sequential adsorption (CSA) using both mean-field theory and numeric simulations. The CSA is defined as a process in which particles are adsorbed onto a line at a rate that depends on the coverage of neighboring sites. Two parameters determine the system dynamics: the equilibrium constant for adsorption/desorption and the cooperativity constant which accounts for interactions with nearest-neighbor particles. In the mean-field model, we derive equations for the time dependence of surface coverage and show that the asymptotic behavior is governed by a second-order phase transition from zero to full coverage. Furthermore, we derive a kinetic equation which allows for a Monte Carlo realization. The Monte Carlo simulation results are in excellent agreement with the analytic solution of the mean-field equations.</p><p><a href="https://reader.elsevier.com/reader/sd/pii/0024379594000395?token=EB0AA78D59A9648480596F018EFB72E0A02FD5FA70326B24B9D501E1A6869FE72CC4D97FA9ACC8BAB56060D6C908EC83">Read the paper</a></p> https://paperswelove.org/papers/customizable-and-extensible-deployment-for-mobilee-a83593a3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Binyu Zang, Enhong Chen, Guoxin Liu, Haibo Chen, Haoran Li, Kang Chen, Qun Huang, Yang Li, Yiming Zhang, Yong Jiang, Yongqiang Xiong <h3>Abstract</h3><p>Mobile and edge-cloud applications are receiving increasing attention due to their ability to provide resourceful and high-quality services for mobile users. However, to optimize mobile applications on mobile and edge-clouds, we need to customize their deployments according to the needs of each application. In this paper, we analyze a set of mobile applications and find that different resources including bandwidth, latency, and computing power critically affect their performance. To better support mobile applications on mobile and edge-clouds, we present CoDe, a customizable and extensible deployment tool for mobile edge-cloud applications, which adopts an agent-based architecture to enable microservice isolation and rapid prototyping. Our evaluation shows that CoDe can effectively improve the deployment flexibility, reduce latency and achieve high microservice isolation while maintaining a negligible performance cost.</p><p><a href="https://www.usenix.org/conference/atc18/presentation/zhang-yiming">Read the paper</a></p> https://paperswelove.org/papers/human-centered-tools-for-coping-with-imperfect-alg-a44da27f/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adrian Weller, Isabel Valera, Krishna P. Gummadi, Manuel Gomez Rodriguez, Muhammad Bilal Zafar <h3>Abstract</h3><p>There is growing interest in using machine learning algorithms to assist experts in application areas ranging from criminal justice to medicine. But as we look to integrate such systems into human decision-making processes, we must ask how people respond to imperfect algorithmic advice in high-stakes situations. We need to carefully design interfaces that enable human decision makers to scrutinize their algorithmic counterparts and correct their errors. In this paper, we study the case of experts (medical professionals) using machine learning algorithms to make predictions. Specifically, we ask whether such experts face cognitive difficulty in adjusting their decisions in response to imperfect algorithmic advice. We develop a simple model of human decision-making when faced with imperfect algorithmic prediction. Based on our model, we run a series of experiments with medical professionals, which reveal that systematically adjusting the algorithms to maximize decision performance is not trivial. Lastly, we offer design recommendations for decision-making tools that can improve individuals' ability to account for algorithmic imperfection.</p><p><a href="https://dl.acm.org/citation.cfm?id=3132763">Read the paper</a></p> https://paperswelove.org/papers/concurrent-cycle-collection-in-reference-counted-s-a3b0a081/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Cormac Flanagan, David F. Bacon, Ron Cytron <h3>Abstract</h3><p>We present a new concurrent cycle collection algorithm for reference counted systems. Existing algorithms cannot insert collection steps concurrently between arbitrary program steps or require a stop-the-world pause to atomicize the marking phase of collection. Our new algorithm, based on the trial deletion method of cycle discovery, allows the object graph of collection candidates to evolve during cycle detection and permits all operations to be performed concurrently, without locks or read barriers. This makes it suitable for real-time and interactive systems. Our algorithm also introduces a cooperative deferred update mechanism which maintains collection invariants by letting each program thread update only the data structures that are private to it.</p><p><a href="https://web.eecs.umich.edu/~weimerw/2012-4610/reading/bacon-garbage.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-internet-census-2012-dataset-an-ethical-examin-a3525a10/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Brett Jordan, Clayton O'Neill, Jeremy Martin, Kevin Butler, Matt Blaze <h3>Abstract</h3><p>In 2012, an anonymous researcher released a dataset outlining the results of a rogue internet census taken between March and December of 2012. This data was collected using unsecured devices discovered using the same probing mechanisms used by the infamous Carna Botnet, which operated using default telnet passwords and targeted over 420,000 devices (many of which were consumer routers). The author provides an analysis of the situation, beginning with a look at the data itself, and moves on to a discussion of ethical issues involved with both the collection of such data and its potential use.</p><p><a href="https://www.researchgate.net/publication/279069631_The_Internet_Census_2012_Dataset_An_Ethical_Examination">Read the paper</a></p> https://paperswelove.org/papers/adam-a-method-for-stochastic-optimization-a33aa2e2/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Diederik P. Kingma, Jimmy Ba <h3>Abstract</h3><p>We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirement, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretation and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide empirical results that demonstrate the method’s state-of-the-art performance.</p><p><a href="https://arxiv.org/pdf/1503.02531.pdf">Read the paper</a></p> https://paperswelove.org/papers/s20-towards-instrumented-data-centers-with-softwar-a2336cd7/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Paul Geissler, Ralf Röhrig <h3>Abstract</h3><p>A challenge facing data centers is migrating existing physical network infrastructures to the improved capabilities of software-defined networking (SDN). Such a migration should take place with minimal disruptions, yet also quickly provide the network operator with better capabilities. SDN makes it possible to instrument data centers with new applications enabling cross-layer solutions and centralized management. This article outlines the basic ideas and approaches towards instrumented data centers using SDN to optimize applications requiring stringent quality-of-service (QoS) levels for data center networking. The proposed approach equips the data center with necessary logic for instrumentation and improvement cycle. It first introduces advantages of SDN for data centers and presents a continuous improvement cycle for advancing data center applications.</p><p><a href="https://ieeexplore.ieee.org/document/6969511">Read the paper</a></p> https://paperswelove.org/papers/clickos-and-the-art-of-network-function-virtualiza-9ff1615d/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 H. Jonathan Chao, Pedro Henrique Penna Martins, Richard Han, Roch Guerin <h3>Abstract</h3><p>Network function virtualization (NFV) is a key component of future networks, aimed at enabling the deployment of network functions as software components on general-purpose hardware, thereby increasing flexibility and reducing costs. In this paper, we explore an extreme design point for NFV where minimalistic virtual machines (VMs) hosting only network functions can boot in less than 30 milliseconds and forward traffic at rates of up to 10 Gb/s. We present the design, implementation, and evaluation of ClickOS, a high-performance virtualization layer, and demonstrate its effectiveness in this extremely low-latency setting.</p><p><a href="https://www.usenix.org/system/files/conference/nsdi14/nsdi14-paper-martins.pdf">Read the paper</a></p> https://paperswelove.org/papers/ix-a-protected-dataplane-operating-system-for-high-9d55873e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adam Belay, Ana Klimovic, Edouard Bugnion, George Prekas, Marton Volk, Mothy Roscoe, Steven Grosbøl <h3>Abstract</h3><p>Modern datacenter applications demand high network throughput while maintaining low latency and providing scaling to many cores. The IX operating system addresses these requirements by restructuring the kernel into a control and dataplane, allowing applications to efficiently access hardware resources. This paper investigates the design and implementation of IX, exploring how it reduces latency and increases throughput in scenario with high request rates. The IX architecture protects application traffic, isolates applications from each other, and supports a large number of concurrent connections. It relies on modern hardware features to offer protection and virtualization typically found in datacenters.</p><p><a href="https://www.usenix.org/system/files/conference/osdi14/osdi14-paper-belay.pdf">Read the paper</a></p> https://paperswelove.org/papers/topology-of-numbers-9975eeb3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Allen Hatcher <h3>Abstract</h3><p>This book presents the basic concepts and results in algebraic topology from the point of view of the theory of numbers. It includes an introduction to the subject, the classical theory of quadratic forms over the integers, as well as lattices, extensively studied in the last few decades.</p><p><a href="https://pi.math.cornell.edu/~hatcher/TN/TNbook.pdf">Read the paper</a></p> https://paperswelove.org/papers/on-the-resemblance-and-containment-of-documents-973f823a/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Andrei Z. Broder <h3>Abstract</h3><p>We develop the mathematical foundations for the analysis of the resemblance and containment of documents as used, for example, in the AltaVista web search engine. We use a simple and efficient fingerprinting scheme introduced by Manber for this process. More generally, we investigate four functions: resemblance and containment (for estimating the similarity of pairs of documents) and max and min (which allow in particular for very accurate estimates of document size). We provide the theoretical underpinnings for the particular implementations used for finding and resolving near-duplicates in a very large scale application. We combine techniques from analysis, probability theory, and linear algebra in order to do this.</p><p><a href="https://www.cs.princeton.edu/courses/archive/spr05/cos598E/bib/broder97resemblance.pdf">Read the paper</a></p> https://paperswelove.org/papers/and-then-there-were-none-91a94aba/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 George V. Neville-Neil <h3>Abstract</h3><p>What happens when we make ourselves obsolete? Consider the Nazi Enigma cipher machine, used with nearly absolute confidence by the German military, until its encryption was compromised by a group of civilians in England led by mathematician and logician Alan Turing. With German confidence in the secrecy of their system shattered, they determined the only solution was replacing those who had proven the systems were insecure with technology itself.</p><p><a href="https://cacm.acm.org/magazines/2013/12/169948-and-then-there-were-none/fulltext">Read the paper</a></p> https://paperswelove.org/papers/the-slab-allocator-an-object-caching-kernel-memory-91047b44/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Jeff Bonwick <h3>Abstract</h3><p>Most general-purpose operating systems provide a memory management facility called the kernel memory allocator. Typically the kernel memory allocator provides variable-size memory allocation through a set of zones, each of which satisfies requests for different size and build. This results in poor memory utilization when the memory allocation pattern is not uniform over time. This paper presents a new kernel memory allocator called the slab allocator. The slab allocator improves the efficiency of object-based allocation by caching recently freed objects so that they can be quickly reallocated. The slab allocator includes a number of optimizations aimed at increasing the effective memory bandwidth and utilization, including "coloring" to reduce cache conflicts. We present measurements which show that the slab allocator can improve system performance substantially relative to other strategies. We also show that the concept of "slab allocation" is quite general, applying to a wide variety of resource management problems beyond memory.</p><p><a href="https://www.usenix.org/legacy/publications/library/proceedings/bos94/full_papers/bonwick.a">Read the paper</a></p> https://paperswelove.org/papers/the-slab-allocator-an-object-caching-kernel-memory-8f595ad8/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Jeff Bonwick <h3>Abstract</h3><p>The object-caching kernel memory allocator provides a convenient and efficient way to manage dynamically allocated kernel data objects. In many computer systems, the kernel must perform frequent allocation and freeing of small objects. Conventional allocation strategies employ a pool of fixed-size objects that are divided into "chunks" to be allocated individually, as needed. Such a scheme typically causes excessive fragmentation and other performance problems. Using some simple and well-known techniques, we have implemented a caching memory allocator that is simple and efficient and that eliminates barriers to good performance of these allocators: fragmentation and lack of space for working data structures. The design and implementation are open to any machine with at least 4k bytes of memory and any operating system.</p><p><a href="https://www.usenix.org/legacy/publications/library/proceedings/bos94/bonwick.html">Read the paper</a></p> https://paperswelove.org/papers/dada-a-crowdsourced-data-marketplace-for-machine-l-8eefc9c4/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Angad Koul, Desiree Dicochea, Esteban Saenzi, Fang-Yi Yu, Lise Getoor <h3>Abstract</h3><p>Machine learning (ML) systems are driven by data, but acquiring high-quality data can be a technical, legal, and logistical challenge. We present Dada, a new crowdsourced data marketplace that allows participants to share data for ML. Dada enables users to earn rewards by providing data for models, and uses blockchain technology to ensure transparency and fair allocation of rewards. We discuss Dada’s platform model, key challenges in crowdsourced data marketplaces, and potential paths forward.</p><p><a href="https://arxiv.org/pdf/1904.09954.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-scalable-commutativity-rule-designing-scalable-8bb89b2f/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Austin T. Clements, Eddie Kohler, M. Frans Kaashoek, Nickolai Zeldovich, Robert Tappan Morris <h3>Abstract</h3><p>Scalability is an important goal and a major challenge in software design for multicore processors. Unfortunately, most software today does not scale well: increasing core counts improve performance much slower than predicted by the hardware increase alone. This paper introduces a new design rule, called the scalable commutativity rule, which holds that whenever interface operations commute, implementations can be made to scale. This rule leads to a new, general interest in commutativity as a guiding principle for software design. The paper describes commutativity conditions for a wide range of common system interfaces and creates a new practical framework, called the commutativity design technique, which helps developers apply this principle. The authors support their rule with examples and a prototype tool, called COMMUTER, to check existing software for scalable operations.</p><p><a href="https://people.csail.mit.edu/nickolai/papers/clements-sc.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-on-line-encyclopedia-of-integer-sequences-874aee46/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 N. J. A. Sloane <h3>Abstract</h3><p>The On-Line Encyclopedia of Integer Sequences (or OEIS) is a database of integer sequences that was started in 1996 and now contains over 60,000 sequences. It has become a useful resource for mathematicians and other scientists, with many publications citing OEIS sequences. The current database is maintained by N. J. A. Sloane and was preceded by two books of sequences compiled by the same author.</p><p><a href="http://neilsloane.com/doc/Me109.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-google-file-system-853f06c1/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Howard Gobioff, Sanjay Ghemawat, Shun-Tak Leung <h3>Abstract</h3><p>We have designed and implemented the Google File System, a scalable distributed file system for large distributed data-intensive applications. It provides fault tolerance while running on inexpensive commodity hardware, and it delivers high aggregate performance to a large number of clients. While sharing many of the same goals as previous distributed file systems, our design has been driven by observations of our application workloads and technological environment, both current and anticipated, that reflect a marked departure from some earlier file system assumptions. This has led us to reexamine traditional choices and explore radically different design points. The file system has been in wide use at Google for more than two years now; it is the underlying storage system for many of our production services as well as research and engineering efforts. In this paper, we present the system's design, discuss many of the implementation issues, and trace the lessons we have learned.</p><p><a href="https://web.archive.org/web/20220912024717/https://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.13.9460&rep=rep1&type=pdf">Read the paper</a></p> https://paperswelove.org/papers/monitoring-system-for-wear-and-tear-on-constructio-831220a6/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Matthias Klein <h3>Abstract</h3><p>Machines and production facilities increase productivity and economic efficiency by substituting manual labor work on a large scale. The industry’s burden of having machines repaired or maintained has attracted more attention from consumers. Therefore it makes sense to assure the availability of machinery by taking appropriate measures in time. A substantial contribution to timely intervention is the continuous monitoring of data generated by machines, which gives immediate feedback about errors and helps to prevent breakdowns. In this study we present a system which detects wear and tear on construction machinery based on acquired sensor data.</p><p><a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.471.9242&rep=rep1&type=pdf">Read the paper</a></p> https://paperswelove.org/papers/minicache-efficient-content-delivery-in-the-mobile-82968631/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Anirban Mahanti, Derek Eager, Niklas Carlsson <h3>Abstract</h3><p>This paper introduces Minicache, a framework to enhance content delivery in mobile edge networks. The strategy involves deploying small caches close to clients to store popular content. This minimizes latency and reduces network traffic. Our evaluation shows that Minicache can significantly improve user experience with minimal additional infrastructure costs. We outline a model for the distribution and management of these caches and discuss potential scalability challenges and solutions.</p><p><a href="http://cnp.neclab.eu/projects/minicache/minicache-workshop.pdf">Read the paper</a></p> https://paperswelove.org/papers/end-to-end-arguments-in-system-design-80cd7a3d/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 D.D. Clark, D.P. Reed, J.H. Saltzer <h3>Abstract</h3><p>This paper presents and elaborates the end-to-end argument, an argument that helps guide placement of functions among the modules of a distributed computer system. The argument suggests that functions placed at low levels of a system may be redundant or of little value when compared with the cost of providing them at that low level. Examples where this argument is applied include bit error recovery, security using encryption, and duplicate message suppression. Low-level mechanisms to support these functions are justified only as performance enhancements. Re-examination of the reasoning leads to several examples of low-level function implementation where the end-to-end argument illuminates their consequent strengths and weaknesses.</p><p><a href="https://web.mit.edu/Saltzer/www/publications/endtoend/endtoend.pdf">Read the paper</a></p> https://paperswelove.org/papers/abc-using-tags-to-strengthen-wifi-security-7f37f5d3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Jean-Pierre Hubaux, Srdjan Capkun <h3>Abstract</h3><p>The ubiquity of WiFi-enabled devices in daily life poses significant security challenges due to the inherent vulnerabilities of wireless communication. In this paper, we propose a novel approach, termed "ABC", that utilizes tags to enhance the security of existing WiFi networks. Our methodology focuses on strengthening the association process between devices and access points through user-involved authentication techniques, which leverage physical context and prevent unauthorized access and spoofing attacks. This approach significantly improves the robustness of network associations without requiring extensive modifications to current communication protocols. Detailed experimental evaluations demonstrate the practicability and efficiency of our solution in real-world scenarios, offering a cost-effective and easily deployable security enhancement for existing WiFi infrastructures.</p><p><a href="http://tik-old.ee.ethz.ch/file/ef72343372ca8659a9ae8a98873167c0/TIKReport302.pdf">Read the paper</a></p> https://paperswelove.org/papers/three-dimensional-analysis-of-rotational-particle--79db6ac1/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Alexis Pasqual, Laurent David, Loïc Fillon, Nicolas Rivière <h3>Abstract</h3><p>The paper presents a method for three-dimensional analysis of rotational particle image velocimetry (3D PIV) data. This method is applied to the characterization of a complex fluid flow. The goal is to offer a convenient framework for the identification and analysis of coherent structures within the flow. The method is validated using both synthetic and experimental data. Results are in agreement with the expected behavior of the studied fluid dynamic fields.</p><p><a href="https://hal.inria.fr/hal-01798793/document">Read the paper</a></p> https://paperswelove.org/papers/a-new-storage-architecture-for-consistent-and-dura-79b483c3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 David Labrecque, Swapnil Bhirud, Willy Zwaenepoel <h3>Abstract</h3><p>Emerging byte-addressable non-volatile memory technologies provide an opportunity to rethink storage architectures. This paper introduces NV-Log, a novel logging mechanism for building consistent and durable data structures using non-volatile memories. NV-Log is an application-transparent, high-performance, and easy-to-use logging mechanism. It can significantly improve the performance of systems that rely on logging for consistency and durability while maintaining data integrity and crash consistency. Through extensive experiments, we demonstrate that NV-Log can outperform traditional mechanisms by up to an order of magnitude in several scenarios.</p><p><a href="https://dl.acm.org/citation.cfm?id=2931093">Read the paper</a></p> https://paperswelove.org/papers/xen-and-the-art-of-virtualization-79a858f3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Alex Ho, Andrew Warfield, Boris Dragovic, Ian Pratt, Keir Fraser, Paul Barham, Rolf Neugebauer, Steven Hand, Tim Harris <h3>Abstract</h3><p>Abstract: In this paper, we describe the design, implementation and evaluation of Xen, an x86 virtual machine monitor, developed to run up to 100 commodity operating systems on a single PC. Unlike traditional virtual machine monitors, Xen achieves high performance for both I/O and computation intensive workloads by taking an unusual approach to virtualizing CPU, memory, and I/O devices. The paravirtualized approach is highly efficient, allowing applications running within Xen guest operating systems to run with high performance. We present extensive benchmark results demonstrating Xen's performance on a range of standard computing benchmarks.</p><p><a href="http://www.cl.cam.ac.uk/research/srg/netos/papers/2003-xensosp.pdf">Read the paper</a></p> https://paperswelove.org/papers/towards-a-framework-for-programming-environments-73f7d949/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Samson Abramsky, Satnam Singh, Simon Gay <h3>Abstract</h3><p>To help resolve the 'programming environment crisis', this paper aims to establish a proper scientific and engineering foundation for the construction of programming environments, analogous to the frameworks used in the past to solve a similar crisis for operating systems. We propose general principles and a theoretical framework for this new element of computing infrastructure, focusing on structuring, modularity and languages. The impact of developments in computer systems and languages upon programming environments is considered, as is the reactive nature of modern computing paradigms.</p><p><a href="https://www.cl.cam.ac.uk/~srk31/research/papers/kell13operating.pdf">Read the paper</a></p> https://paperswelove.org/papers/unleashing-the-power-of-unikernels-with-rump-kerne-721f409d/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Antti Kantee <h3>Abstract</h3><p>Recent advances made in cloud computing have made the concept of an operating system, and by extension hardware architecture, less rigid. Even complex protocols and standards are not a prerequisite for all computing environments. This kind of freedom yields opportunities for rethinking the operating system altogether, often combining with virtualization to reduce the operating system down to a set of specialized libraries running directly on top of a hypervisor - referred to as a unikernel. Unikernels have been proposed as a way to reduce the overhead present in traditional operating systems and increase the security by reducing code complexity and attack surface.</p><p><a href="https://www.usenix.org/system/files/login/articles/login_1410_03_kantee.pdf">Read the paper</a></p> https://paperswelove.org/papers/unikernels-library-operating-systems-for-the-cloud-6fb5ec04/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Amir Chaudhry, Anil Madhavapeddy, Balraj Singh, David J. Scott, David Sheets, Ian Leslie, Jeremy Yallop, Jon Crowcroft, Jonathan Ludlam, Lionel Rieg, Magnus Vigfusson, Raphael sharp, Richard Mortier, Steven Smith, Thomas Gazagnaire, Thomas Leonard <h3>Abstract</h3><p>Unikernels are specialized, single-address-space machine images constructed by using library operating systems. Rather than providing a stable functionality in the form of files and profiles, these images are linked with the application source code and configuration file in order to produce a single bootable image that can run on a cloud provider such as Amazon EC2, Xen, or Google Compute Engine. Such images offer advantages over traditional virtual-machine images, including reduced attack surface, improved boot times, and reduced system overhead. In addition, they improve functional density, flexibility, and integration of the entire software stack.</p><p><a href="https://cacm.acm.org/magazines/2014/1/170866-unikernels/fulltext">Read the paper</a></p> https://paperswelove.org/papers/on-the-origin-of-gravity-and-the-laws-of-newton-6e911cd0/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Erik P. Verlinde <h3>Abstract</h3><p>Starting from first principles and basic insights from thermodynamics and holography we derive Newton's laws of motion and gravity by regarding them as emergent phenomena. In particular, the universal nature of gravity is shown to be related to the laws of thermodynamics. We argue that inertia follows from an opposition of the tendency of matter to increase its entropy against the entropy gradients associated with other matter. As in the case of the gravitational force law, also the law of inertia can be derived from general principles such as energy conservation, the equivalence principle and by assuming the entropy formula holds. Our approach thus illuminates the connection between gravity as an entropic force and the thermodynamics of spacetime.</p><p><a href="https://arxiv.org/pdf/1407.4245v1.pdf">Read the paper</a></p> https://paperswelove.org/papers/on-the-security-of-next-generation-data-center-net-6e63ff60/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Eric Keller, Michael Freedman <h3>Abstract</h3><p>The next generation of data center networks, such as those using Software Defined Networking (SDN) and cloud computing technologies, promises to increase efficiency and flexibility while lowering costs. However, they also bring new security challenges such as providing full network isolation, verifying security compliance, and securing centralized controllers from attack. In this work, we outline the new security issues particular to these networks, propose initial solutions and a taxonomy to analyze attacks, all while considering the SDN and cloud deployments currently undergoing experimentation in data centers.</p><p><a href="http://conferences.sigcomm.org/sigcomm/2013/papers/hotsdn/p67.pdf">Read the paper</a></p> https://paperswelove.org/papers/a-simulation-of-register-allocation-using-shortest-6d75cf23/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Blau-Heller, Munemori, Saeedi <h3>Abstract</h3><p>In this paper, we present a novel approach to register allocation called SRAS (Shortest paths Register Allocation Simulation). By using a graph theoretical model, we can describe the register allocation problem in terms of finding shortest paths. We show that this method can be applied to existing compilers and yields comparable or improved results over traditional methods in various cases.</p><p><a href="https://open.library.ubc.ca/media/download/pdf/24/1.0052016/1">Read the paper</a></p> https://paperswelove.org/papers/design-and-implementation-of-a-non-disruptive-upgr-6b8b71cf/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Bruce Chen, Dan Boneh, Mendel Rosenblum, Tal Garfinkel, Terence Kelly <h3>Abstract</h3><p>Today's leading Internet services require high availability and reliability from their server infrastructure. These services also need frequent software upgrades, but upgrading large-scale distributed server systems without causing disruptions is difficult. We present the design, implementation, and evaluation of a non-disruptive upgrade infrastructure for scalable Internet services based on virtual machine technology. Our approach leverages virtual machines (VMs) to provide fault isolation to support non-disruptive software upgrades. Specifically, we introduce a two-level virtual machine design and a hardware abstraction layer allowing servers to apply OS updates, reconfigure services, and recover from faults with minimal downtime.</p><p><a href="http://www.cse.psu.edu/~trj1/cse543-f12/docs/p132-levin-hydra.pdf">Read the paper</a></p> https://paperswelove.org/papers/kernohan-a-high-integrity-high-assurance-microkern-662d183c/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Daryl Seward, Gernot Heiser, Kevin Elphinstone, Philip Derrin <h3>Abstract</h3><p>The proposed presentation introduces Kernohan, a new microkernel designed for high-assurance embedded systems, as part of our work on trustworthy systems. Kernohan’s design builds on our existing high-assurance, formally-verified seL4 kernel. It is intended as a platform for secure, safety-critical software, with a particular focus on defence, automotive, and avionics applications. Kernohan embodies several innovations for supporting static resource management, security, and fault-tolerance. Additionally, the presentation discusses results from our evaluation which demonstrate the performance and integrity benefits of Kernohan in comparison to current state-of-the-art microkernels like L4-embedded and CAmkES.</p><p><a href="http://trustworthy.systems/publications/csiro_full_text/Elphinstone_ZMH_17.pdf">Read the paper</a></p> https://paperswelove.org/papers/clive-a-seamless-environment-for-heterogeneous-sys-64bdb078/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Chris Liedel, Gerald Popek, John Bruno, Richard Guy <h3>Abstract</h3><p>Distributed computing systems present several new problems to users and systems designers alike. As the complexity and number of different machine types and system-level services grow, specific approaches are required to seamlessly support a wide range of machines and programs. CLIVE represents a set of powerful software abstractions to provide uniform access to the services available in a wide collection of heterogeneous computing environments. These abstractions define a seamless environment that masks the complexities inherent in widely diverse system constructions and methodologies. This paper describes the implementation of the CLIVE system, concentrating on the design and architecture of the abstractions employed, as well as some simple examples of their use. We demonstrate the advantages of uniform approaches to file system integration and distributed program execution.</p><p><a href="http://github.com/fjballest/docs/blob/master/clivesys.pdf">Read the paper</a></p> https://paperswelove.org/papers/algebraic-topology-6479ef31/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Allen Hatcher <h3>Abstract</h3><p>This document presents an introduction to some of the fundamental concepts in algebraic topology. Topics covered include the study of homological and cohomological theories, as well as basic homotopy theory and the study of higher homotopy groups.</p><p><a href="https://pi.math.cornell.edu/~hatcher/AT/AT.pdf">Read the paper</a></p> https://paperswelove.org/papers/contiki-a-lightweight-and-flexible-operating-syste-5c78dabe/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adam Dunkels <h3>Abstract</h3><p>Recent advances in microelectronics have made it feasible to develop low-cost, low-power delay-tolerant sensor networks. This paper introduces Contiki, a lightweight and flexible operating system for tiny networked sensors. Contiki is designed to run on small memory devices and allows dynamic loading and unloading of programs. It supports both multi-threading and event-driven programming, making it portable and efficient for a wide range of applications. With Contiki, networked sensors can be set up easily, and the system is designed to be highly modular, making it adaptable to different hardware platforms and application domains.</p><p><a href="http://dunkels.com/adam/dunkels04contiki.pdf">Read the paper</a></p> https://paperswelove.org/papers/what-is-a-young-tableau-5b664ca0/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Alexander Yong <h3>Abstract</h3><p>A Young tableau is a combinatorial object useful in representation theory and algebraic geometry, particularly Schubert calculus. It was first used by Alfred Young in 1900 to study symmetric group representations, but it was not until later in the twentieth century that the tableau became ubiquitous in algebraic combinatorics. In this article, we'll explain what a Young tableau is and highlight some remarkable applications.</p><p><a href="https://www.ams.org/notices/200702/whatis-yong.pdf">Read the paper</a></p> https://paperswelove.org/papers/gender-differences-in-grief-after-the-death-of-a-p-5b35d853/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Alan R. Kemp, Clarissa J. Humphrey <h3>Abstract</h3><p>This study examined gender differences in the experience of grief during the initial six months following the death of a parent. One hundred and seventy-seven adults, who had lost a parent, completed self-administered questionnaires related to demographic characteristics, pre-death characteristics of the parent-offspring relationship, circumstances of the parental death, social support, coping strategies, and grief symptoms. The findings revealed women experienced a higher level of grief symptoms compared to men. Additionally, different predictors of grief symptoms were identified for men and women. These differences are discussed in terms of sociocultural perspectives and implications for counseling.</p><p><a href="https://osf.io/6qzxc/">Read the paper</a></p> https://paperswelove.org/papers/eat-chapter-9-583a4f68/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Robert Ghrist <h3>Abstract</h3><p>This chapter comprises an introduction to persistent homology, a central parameterization of scale for topological data analysis. The primary application of persistence, the topological analysis of coverage in sensor networks, is detailed. Along the way, the requisite ideas from applied topology are introduced, including homology and various representations of persistence.</p><p><a href="https://www.math.upenn.edu/~ghrist/EAT/EATchapter9.pdf">Read the paper</a></p> https://paperswelove.org/papers/composable-and-compilable-macros-you-want-it-when-561b4f89/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Matthias Felleisen, Ryan Culpepper <h3>Abstract</h3><p>We present a macro system that supports a disciplined style of macro use based on the idea of composable and compilable macros. In this style, programmers design and implement components with macros using an API consisting only of functions. This principled approach enables a design-by-contract methodology of assuming specific properties of arguments to macros. We demonstrate this approach with a module system that ensures both syntactic and semantic correctness. We also present a novel method for optimizing compile-time dependencies.</p><p><a href="http://www.ccs.neu.edu/racket/pubs/icfp10-cf.pdf">Read the paper</a></p> https://paperswelove.org/papers/k-anonymity-a-model-for-protecting-privacy-524971b2/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Latanya Sweeney <h3>Abstract</h3><p>We consider the problem of linking de-identified information such as diagnosis, gender and date of birth to identified information such as a name to uniquely and completely determine identity. We demonstrate that data in medical, as well as non-medical, databases can be linked by knowing only a small amount of information about an individual. We introduce k-anonymity as a way to protect individual privacy and describe a formal protection model.</p><p><a href="https://epic.org/wp-content/uploads/privacy/reidentification/Sweeney_Article.pdf">Read the paper</a></p> https://paperswelove.org/papers/fine-grained-contention-management-for-concurrent--51bb4d3c/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Daisuke Takahashi, Gabor Gerofi, Kazutomo Yoshii, Yutaka Ishikawa <h3>Abstract</h3><p>As computational concurrency in high-performance computing (HPC) systems increases, it becomes critical to manage concurrent control operations at a fine-grained level to avoid performance bottlenecks. This paper proposes a novel mechanism for fine-grained contention management that enhances the performance of concurrent control operations in operating systems. The proposed approach is evaluated through various benchmark experiments demonstrating substantial improvements in system throughput and latency.</p><p><a href="http://www.mcs.anl.gov/events/workshops/ross/2015/slides/ross2015-gerofi.pdf">Read the paper</a></p> https://paperswelove.org/papers/statecharts-a-visual-formalism-for-complex-systems-46a2275a/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 David Harel <h3>Abstract</h3><p>A broad extension of the conventional formalism of state machines and state diagrams, relevant to the specification and design of complex discrete-event systems, such as multi-computer real-time systems, communication protocols, and digital control units. The basic idea is to replace the timing element of state machines with a set of control states that relate to the system being modelled.</p><p><a href="https://www.weizmann.ac.il/math/harel/sites/math.harel/files/users/user56/Statecharts.pdf">Read the paper</a></p> https://paperswelove.org/papers/barrelfish-a-multikernel-os-for-multicore-hardware-46745895/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Abraham Welc, Andrew Baumann, Paul Barham, Pierre-Evarist Gaudry, Reto Schüpbach, Timothy Harris <h3>Abstract</h3><p>Recent trends in computer architecture have radically changed the landscape of systems software. Multicore processors, non-uniform memory access architectures, and the growing importance of accelerators mean that OS developers face the challenge of building systems for hardware whose characteristics vary widely. Barrelfish is a new operating system designed from the ground up for scalability through increased use of message passing and more extensive use of distributed data structures in the OS. In this paper, we describe the rationale behind the Barrelfish architecture, and the detailed design of the system and implementation of core OS components, we also show early results evaluating the system.</p><p><a href="http://www.barrelfish.org/publications/barrelfish_mmcs08.pdf">Read the paper</a></p> https://paperswelove.org/papers/design-and-implementation-of-a-data-extraction-too-452f7bb5/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Gilbert Chu, John R. Barry <h3>Abstract</h3><p>Data extraction from complex and multifaceted documents, like those used in refractory design, poses significant challenges in document interpretation, structuring, and storage. This paper outlines the architecture and implementation of a data extraction tool designed specifically to handle the intricacies of refractory design documents. The tool automates the conversion of unstructured document content into a format that can be easily integrated into a database management system. It enables users to query design specifications and generate reports without needing to manually parse through extensive documentation.</p><p><a href="http://people.csail.mit.edu/jrb/Projects/dexprs.pdf">Read the paper</a></p> https://paperswelove.org/papers/cloud-networking-challenges-and-opportunities-for--446653d1/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Chadi Assi, Mohammad A. Salahuddin, Yahia A. Samhan, Yaser Ghanem <h3>Abstract</h3><p>The rapid growth of cloud networking has brought enormous benefits to society. However, such growth has also introduced several challenges that need to be addressed in order to harness its full potentials. This paper explores various cloud networking issues including security, energy efficiency, and monitoring, which are important pillars for realizing innovations in this realm. We discuss some of the challenges facing cloud networking and identify key enablers for innovations that will foster the development of future solutions.</p><p><a href="http://aura.abdn.ac.uk/bitstream/handle/2164/7988/PID4488667.pdf?sequence=1&isAllowed=y">Read the paper</a></p> https://paperswelove.org/papers/hotos-jeremiad-41f4e51e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Donald E. Porter, Scott L. Donaldson, Yuanyuan Zhou <h3>Abstract</h3><p>The paper discusses the prevalence of software crashes and the importance of designing systems that are robust against such failures. It presents insights into the challenges of improving system reliability and suggests approaches to reduce the rate of software-induced downtime through better design and engineering practices.</p><p><a href="http://studies.ac.upc.es/doctorat/ENGRAP/hotos-jeremiad.pdf">Read the paper</a></p> https://paperswelove.org/papers/privacy-accuracy-and-consistency-a-data-analysis-o-4176b9af/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Arvind Narayanan, Vitaly Shmatikov <h3>Abstract</h3><p>We present a new class of privacy attacks against high-dimensional micro-data, based on the observation that large-scale data analyses often produce aggregate results that are available to the adversary. As a case study, we demonstrate our attacks on the Netflix Prize dataset, which contains anonymous movie ratings for over 500,000 subscribers, released as part of an open competition to improve Netflix’s movie recommendation algorithm. We show that an adversary who knows only a little bit about an individual subscriber can easily identify this subscriber’s record if it is present in the dataset. Our results highlight the risks of anonymized data, demonstrating that even heavily “anonymized” datasets are not immune to privacy attacks.</p><p><a href="https://www.cs.utexas.edu/~shmat/shmat_oak08netflix.pdf">Read the paper</a></p> https://paperswelove.org/papers/the-same-origin-policy-3c754eb1/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Adam Barth, Collin Jackson, John C. Mitchell <h3>Abstract</h3><p>The same-origin policy is a critical security mechanism that restricts how documents or scripts loaded from different origins can interact. The goal of this paper is to provide a comprehensive understanding of the same-origin policy, how it works in practice, and the various ways in which it is enforced by different browsers, including inconsistencies and loopholes that can be exploited by attackers. We also survey a variety of methods and proposals designed to enhance or circumvent the policy, such as cross-origin resource sharing, postMessage, JSONP, and Flash.</p><p><a href="https://crypto.stanford.edu/sameorigin/sameorigin.pdf">Read the paper</a></p> https://paperswelove.org/papers/investigating-user-behavior-in-home-computer-secur-397d4936/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Anna Oleszkiewicz, Jae Park, Yang Wang <h3>Abstract</h3><p>This study investigates how private home computer users handle their computer security. The focus lies on user behavior regarding home computer security, including actions they take to protect their systems, reasons behind these actions, and perceptions regarding threats and vulnerabilities. Our study is based on data gathered from a survey distributed to a sample of home users. The results provide an insight into user practices and attitudes towards different aspects of computer security, the perceived importance of these aspects, as well as users' security expertise and need for education. This study further highlights several factors influencing user behavior and its implications for the design of future computer security tools and educational initiatives.</p><p><a href="https://www.sciencedirect.com/science/article/pii/S016740480900073X">Read the paper</a></p> https://paperswelove.org/papers/deep-eigenspace-estimation-of-3d-brain-activation--37d2f2ee/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Brett Larsen, David G. Stork <h3>Abstract</h3><p>We propose a novel technique for the estimation of spatial activation patterns in functional neuroimaging data, cast as an inverse spatial eigenspace estimation problem. Previous methods include regression-based approaches which are slow and sensitive to noise and assumptions of Gaussianity. By using deep learning, our proposed method, DEEP, trains an autoencoder neural network to map from signals to brain activation maps that minimize an eigenspace estimation loss function. Simulated experiments demonstrate the applicability and utility of our method in reducing the dimensionality of complex brain imaging data and improving estimation of brain activation patterns.</p><p><a href="https://arxiv.org/pdf/1802.04914.pdf">Read the paper</a></p> https://paperswelove.org/papers/kinetic-sand-topics-in-cloud-scale-data-rebalancin-3756140b/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Christopher Stewart, Trey Gordon Williams, Walter Rivera <h3>Abstract</h3><p>Data rebalancing is essential to maintaining scalability in cloud storage systems. When more storage space becomes available, the cloud must move data to reconstruct balanced distributions over load and capacity. However, existing rebalancing methods spend too much time observing system state and incur alternative costs while waiting for this state information. In this paper, we propose Kinetic Sand, a method that rebalances data in cloud storage systems without observing system state. Instead, Kinetic Sand achieves fairness while content is migrated. We present a series of simulation results to demonstrate the fairness and cost efficiency of Kinetic Sand.</p><p><a href="https://www.usenix.org/system/files/conference/hotcloud16/hotcloud16_williams.pdf">Read the paper</a></p> https://paperswelove.org/papers/minuet-rethinking-concurrency-control-in-storage-s-365b9beb/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Anastasia Ailamaki, Donald Kossmann, Gregor Alonso, Oliver Zellweger, Tim Kraska <h3>Abstract</h3><p>Existing implementations of concurrency control, such as two-phase locking or optimistic concurrency control, offer only a fixed trade-off between performance and isolation. Two-phase locking provides high isolation but poor performance scalability. Optimistic concurrency control conversely provides better performance scalability but may have issues with correctness when transactions conflict. In this paper, we propose Minuet, a new form of concurrency control, as a scalable and correct alternative to these existing protocols. Minuet is a multi-version scheme inspired by software transactional memory. It operates under a straightforward property: each transaction executes under a snapshot view of the database from when the transaction started, and visible updates from other transactions are applied more quickly than waiting for the slower durability updates. We show that Minuet addresses the scalability limitations found in existing protocols and interacts correctly with all the necessary durability restrictions from current standards, giving transaction isolation comparable to that seen with two-phase locking, while preserving performance scalability.</p><p><a href="https://www.usenix.org/system/files/conference/osdi14/osdi14-paper-zellweger.pdf">Read the paper</a></p> https://paperswelove.org/papers/on-compact-sets-in-horn-logic-34c6747d/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Francesco M. Donini, Marco Cadoli, Marco Schaerf, Paola Liberatore <h3>Abstract</h3><p>Horn logic is a central topic in artificial intelligence. While it is usually employed to generate correct conclusions from true hypotheses, this paper focuses on the complementary question: how to derive all hypotheses that entail a given set of conclusions. This task can be succinctly encoded using the notion of compact sets, which are pairs of sets (hypotheses, conclusions) that are equivalent (that is, have the same models), but where no pair exists with either hypotheses or conclusions being proper subsets of the corresponding element of the pair. This notion is characterized according to several dimensions: logical form of conclusions (definite clauses, Horn clauses, or propositions), possible size of conclusions (multiple conclusions, a single conclusion, no trivial conclusions), possible size of hypotheses (empty, everything), and presence of disjunction. A computational analysis is also given considering both the generation and recognition problem for definite propositional Horn clauses.</p><p><a href="http://arxiv.org/abs/1004.4240">Read the paper</a></p> https://paperswelove.org/papers/how-to-evaluate-the-performance-of-your-web-site-t-33dd0b8a/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Fabian E. Bustamante, Y. Charlie Hu <h3>Abstract</h3><p>The increasing importance of the Web has made the performance of Web sites a crucial concern, both for end users and for service providers. In recent years the Internet research community has developed server-side, network-based, and client-side approaches to evaluate the performance of a Web site. In this paper, we argue that none of these approaches alone suffices to fully understand and characterize the performance of all parts involved in the Web transaction. We present and discuss a Web Timing Framework based on the analysis of recent work, which for the first time offers an integral approach to top-down Web performance evaluation. This framework is flexible enough to incorporate future research results. We also present WebTP, a tool based on our framework, which allows the seamless integration and correlation of results in a user-friendly graphical environment.</p><p><a href="https://www.cs.jhu.edu/~fabian/courses/CS600.424/course_papers/webtiming.pdf">Read the paper</a></p> https://paperswelove.org/papers/a-virtual-machine-introspection-based-architecture-30860a28/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Dawn Song, Xuxian Jiang <h3>Abstract</h3><p>Virtual Machine Monitors (VMMs) have become increasingly popular as a platform for building secure systems. In this paper, we present a novel IDS architecture based on VMM technology — introducing a new security layer below operating systems to provide improved security. This architecture permits the implementation of intrusion detection services without using tampering techniques that run in the target system and interfere with its normal operation. By residing in the VMM, our architecture can inspect the state of a virtual machine externally and leverage the VMM's isolation properties to remain protected from potential attacks. Our VMM-based IDS infrastructure provides performant and robust intrusion monitoring capabilities, which we argue is a vital step forward for intrusion detection technologies.</p><p><a href="http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.117.6702&rep=rep1&type=pdf">Read the paper</a></p> https://paperswelove.org/papers/semi-supervised-sequence-learning-2ea25fdf/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Andrew M. Dai, Quoc V. Le <h3>Abstract</h3><p>Deep Neural Networks (DNNs) are powerful models that have achieved excellent performance on difficult learning tasks. Although DNNs work well whenever large labeled training sets are available, they are less effective in scenarios where few labeled data points are accessible. To address this, we present the Sequence Autoencoder (SAE) – a generalization of the autoencoder to sequence input and sequence output such that the input, a sequence x, is encoded to a hidden representation which is then decoded to reconstruct the original sequence. We show that models learned with the SAE can be used to define an unsupervised learning algorithm for sequence prediction tasks, in which a sequence y is predicted given sequence x. This unsupervised pretraining algorithm works anytime the input/output can be cast to a sequence prediction problem, including language modeling and text classification. We demonstrate the effectiveness of the Sequence Autoencoder by applying it to semi-supervised sentiment classification, where the model achieves state-of-the-art results. Furthermore, we show how to augment the SAE with the Long Short-Term Memory (LSTM) networks, which helps the model better capture long-distance dependencies in the data. Experiments demonstrate that our approach consistently outperforms competitive baselines on a language modeling task and a semi-supervised sentiment classification task.</p><p><a href="https://arxiv.org/pdf/1605.06285.pdf">Read the paper</a></p> https://paperswelove.org/papers/time-provisioning-evaluation-of-kvm-docker-and-uni-2e027055/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Christian Westphal, Leonardo Karnopp, Lisandro Zambenedetti Granville, Luiz Fernando Bittencourt <h3>Abstract</h3><p>The use of cloud computing provides elasticity by dynamically leveraging the efficient use of allocated resources. To effectively achieve elasticity, resources on cloud platforms must be provisioned at a fast pace. In this work, we perform an evaluation of provisioning times of three virtualization technologies, Kernel-based Virtual Machine (KVM), Docker, and Unikernels, in terms of performance and efficiency on cloud platforms. Our results show significant differences among these technologies considering the infrastructure's performance requirements.</p><p><a href="http://repositorio.pucrs.br/dspace/bitstream/10923/14178/2/Time_provisioning_evaluation_of_KVM_Docker_and_Unikernels_in_a_Cloud_Platform.pdf">Read the paper</a></p> https://paperswelove.org/papers/high-performance-multi-channel-mpi-alltoall-2ca8f1ea/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Bharat Kumar, Dhabaleswar K. Panda, Jiuxing Liu, Wei Huang <h3>Abstract</h3><p>The MPI_Alltoall collective is used in a variety of parallel scientific applications. In this paper, we identify the scalability problem of the MPI_Alltoall and provide a novel Multi-Channel approach to improving the performance characteristics, including latency and bandwidth of this important collective operation.</p><p><a href="http://mvapich.cse.ohio-state.edu/static/media/publications/abstract/huangwei-ics06.pdf">Read the paper</a></p> https://paperswelove.org/papers/meltdown-reading-kernel-memory-from-user-space-2a47912d/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Anders Fogh, Daniel Genkin, Daniel Gruss, Jann Horn, Michael Schwarz, Mike Hamburg, Moritz Lipp, Paul Kocher, Stefan Mangard, Thomas Prescher, Werner Haas, Yuval Yarom <h3>Abstract</h3><p>Meltdown is a novel attack that allows overcoming memory isolation and reading arbitrary kernel-memory location, including personal data and passwords, using a thirdparty program. Meltdown exploits side effects of out-of-order execution on modern processors to read arbitrary kernelmemory locations. As a consequence, personal data can be stolen from the affected system. Meltdown allows attackers to circumvent the separation between user applications and the operating system, affecting multiple operating systems. Moreover, the attack does not only work on personal computers but can also be exploited in the cloud.</p><p><a href="https://meltdownattack.com/meltdown.pdf">Read the paper</a></p> https://paperswelove.org/papers/live-updating-unikernels-29559936/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 David Chisnall, Malte Schwarzkopf, Simon W. Moore, Stephen Kell, Thomas Joergersen <h3>Abstract</h3><p>Unikernels are low-footprint virtual machines, often written in low-level or unmanaged programmed languages such as C or OCaml. They are used to construct cloud services due to their fast boot times, low resource footprints and strong isolation. Today, unikernel systems require stopping the server to perform software updates, which harms availability. Solutions such as migrating pre-forked processes have been explored, but these are unsuitable for unikernels. We present a framework for live-updating unikernels, which combines compile- and link-time feedback with a low-overhead migration mechanism. We demonstrate that this approach can achieve downtime on par with process live-patching, while retaining the resource savings of unikernels. Our approach is application-agnostic, supporting a range of unikernel applications with no code changes required, and is demonstrated with MirageOS and Rumprun.</p><p><a href="https://www.duo.uio.no/bitstream/handle/10852/59240/45/live-updating-unikernels.pdf">Read the paper</a></p> https://paperswelove.org/papers/ballista-a-fault-injection-based-automated-testing-24a35732/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Darren Heimbigner, John DeVale, Philip J. Koopman, Shing chi Cheung, Williy E. Hasselbring <h3>Abstract</h3><p>Ballista is an experimental testing service designed for blind, automated software testing to explore the dependability limit of application programs via fault injection. Ballista aims to scale fault injection testing methods to provide an unbiased foundation for specification-based testing of commercial off-the-shelf (COTS) software as well as custom applications. We outline a scalable validation framework, describe our prototype implementation for testing ANSI C POSIX APIs, and present validation results that demonstrate the applicability and limitations of service-oriented fault injection.</p><p><a href="https://users.ece.cmu.edu/~koopman/ballista/ftcs99/ftcs99.pdf">Read the paper</a></p> https://paperswelove.org/papers/exokernel-an-operating-system-architecture-for-app-2049bf3b/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Dawson R. Engler, James O'Toole, M. Frans Kaashoek <h3>Abstract</h3><p>Operating systems must currently manage machine resources to provide appropriate and controlled sharing among applications, as well as to protect applications from each other. However, this software resource management has proven time consuming and helps to impose inefficiencies on applications. In this paper, we present a new operating system architecture, exokernel, that safely multiplexes available hardware resources while respecting application-specific customization of traditional operating system abstractions. We have implemented an exokernel-based system for 32-bit x86 computers. Measurements show that traditional operating system abstractions can be implemented efficiently at application level by using the low-level secure multiplexing mechanisms provided by an exokernel. The results show that the exokernel’s secure multiplexing of hardware resources can enable application-specific customization to improve performance and flexibility with no loss of the protection provided by traditional operating systems.</p><p><a href="http://pages.cs.wisc.edu/~remzi/Classes/736/Spring2014/Papers/exo-sosp95.pdf">Read the paper</a></p> https://paperswelove.org/papers/dive-into-deep-learning-1e07bb57/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Alexander J. Smola, Aston Zhang, Mu Li, Zachary C. Lipton <h3>Abstract</h3><p>This open-source book represents our attempt to make deep learning approachable, teaching students what works, what doesn’t, and how to reason about deep learning methods. We provide full chapters on modern codes and concept explanations on a breadth of topics in deep learning, including experimental design, ethical considerations, applications, and more.</p><p><a href="https://arxiv.org/pdf/1809.07701.pdf">Read the paper</a></p> https://paperswelove.org/papers/osvoptimizing-the-operating-system-for-virtual-mac-1a85664e/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Avi Kivity, Christoph Hellwig, Dor Laor, Eduardo Habkost, Martin Rotschild <h3>Abstract</h3><p>In the last decade, virtualization has gone from being a promising technology to the basis of most data centers. However, today’s virtual machines use the same operating systems as physical machines—operating systems that were designed to share hardware between mutually-untrusting applications. These traditional operating systems are less than optimal when deployed in a virtual machine, causing performance and efficiency issues. OSv is a new operating system designed specifically for running a single application on a virtual machine. By rethinking the traditional interface between the operating system kernel, virtual machine monitor, and application, OSv eliminates many of the sources of inefficiency and complexity present in traditional operating systems. Novel techniques in OS design are proposed to significantly improve performance in key workloads while retaining the strong isolation across virtual machines and easy application porting that virtualization provides. This paper explores these trade-offs and presents an experimental evaluation of OSv running key workloads on KVM over IBM’s Cloud Workload Factory.</p><p><a href="https://www.usenix.org/system/files/conference/atc14/atc14-paper-kivity.pdf">Read the paper</a></p> https://paperswelove.org/papers/a-persistent-system-in-real-use-experiences-of-the-17db25b3/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Jochen Liedtke <h3>Abstract</h3><p>In 1979, a persistent multi-user operating system was installed and made available to the computer science department at the University of Bielefeld. This paper reports on the experiences gained during its 13-year usage. The OS is based on segmented address spaces and capability-based protection. Its persistent virtual memory is managed by a garbage collector running as a background process. High-level language programs can be mapped directly into persistent memory which is demand-paged and backed up on disk. The persistent system allows for the transparent migration of complete yet small virtual machines (about 20 KBytes per user) between different computers connected by a LAN. This proved useful not only for balancing load among computers but also for fault tolerance and user mobility. The application of persistent memory in real-time environments will also be discussed, as well as a persistent object system, an optimized garbage collector for implementation of bounded response time, and the lessons learned from its longtime use.</p><p><a href="https://os.itec.kit.edu/downloads/publ_1993_liedtke_persistent-system-in-real-use.pdf">Read the paper</a></p> https://paperswelove.org/papers/broken-promises-of-privacy-responding-to-the-surpr-16dd0859/ 2026-01-24T23:47:00+00:00 2026-01-24T23:47:00+00:00 Paul Ohm <h3>Abstract</h3><p>This article critiques the failure of reidentification attacks and their implausibility assumptions that drive the anonymization. It aims to explore a regulatory response to reidentification harms as it also urges a shift to protecting data privacy contextually.</p><p><a href="https://www.uclalawreview.org/broken-promises-of-privacy-responding-to-the-surprising-failure-of-anonymization-2/">Read the paper</a></p>