Search

We develop and evaluate a structure learning algorithm for clinical time series. Clinical time series are multivariate time series observed in multiple patients and irregularly sampled, challenging existing structure learning algorithms. We assume that our times series are realizations of StructGP, a k-dimensional multi-output or multi-task stationary Gaussian process (GP), with independent patients sharing the same covariance function. StructGP encodes ordered conditional relations between time series, represented in a directed acyclic graph. We implement an adapted NOTEARS algorithm, which based on a differentiable definition of acyclicity, recovers the graph by solving a series of continuous optimization problems. Simulation results show that up to mean degree 3 and 20 tasks, we reach a median recall of 0.93% [IQR, 0.86, 0.97] while keeping a median precision of 0.71% [0.57-0.84], for recovering directed edges. We further show that the regularization path is key to identifying the graph. With StructGP, we proposed a model of time series dependencies, that flexibly adapt to different time series regularity, while enabling us to learn these dependencies from observations.

We present Union, a trust-minimized bridge protocol that enables secure transfer of BTC between Bitcoin and a secondary blockchain. The growing ecosystem of blockchain systems built around Bitcoin has created a pressing need for secure and efficient bridges to transfer BTC between networks while preserving Bitcoin's security guarantees. Union employs a multi-party variant of BitVMX, an optimistic proving system on Bitcoin, to create a bridge that operates securely under the assumption that at least one participant remains honest. This 1-of-n honest approach is strikingly different from the conventional honest-majority assumption adopted by practically all federated systems. The protocol introduces several innovations: a packet-based architecture that allows security bonds to be reused for multiple bridge operations, improving capital efficiency; a system of enablers to manage functionaries participation and to enforce penalties; a flexible light client framework adaptable to various blockchain architectures; and an efficient stop watch mechanism to optimize time-lock management. Union is a practical and scalable solution for Bitcoin interoperability that maintains strong security guarantees and minimizes trust assumptions.

The well-known problem stated by A. Meir and L. Moser consists in tiling the unit square with rectangles (details), whose side lengths equal $\frac1n\times\frac1{n+1}$, where indices $n$ range from 1 to infinity. Recently, Terence Tao has proved that it is possible to tile with $\bigl(\frac1n\bigr)^t\times\bigl(\frac1{n+1}\bigr)^t$ rectangles (squares with the side length of $\bigl(\frac1n\bigr)^t$), $1/2

Formal verification using proof assistants, such as Coq, enables the creation of high-quality software. However, the verification process requires significant expertise and manual effort to write proofs. Recent work has explored automating proof synthesis using machine learning and large language models (LLMs). This work has shown that identifying relevant premises, such as lemmas and definitions, can aid synthesis. We present Rango, a fully automated proof synthesis tool for Coq that automatically identifies relevant premises and also similar proofs from the current project and uses them during synthesis. Rango uses retrieval augmentation at every step of the proof to automatically determine which proofs and premises to include in the context of its fine-tuned LLM. In this way, Rango adapts to the project and to the evolving state of the proof. We create a new dataset, CoqStoq, of 2,226 open-source Coq projects and 196,929 theorems from GitHub, which includes both training data and a curated evaluation benchmark of well-maintained projects. On this benchmark, Rango synthesizes proofs for 32.0% of the theorems, which is 29% more theorems than the prior state-of-the-art tool Tactician. Our evaluation also shows that Rango adding relevant proofs to its context leads to a 47% increase in the number of theorems proven., Comment: In Proceedings of the 47th International Conference on Software Engineering (ICSE), Ottawa, ON, Canada, April 2025

Software engineers frequently grapple with the challenge of accessing disparate documentation and telemetry data, including Troubleshooting Guides (TSGs), incident reports, code repositories, and various internal tools developed by multiple stakeholders. While on-call duties are inevitable, incident resolution becomes even more daunting due to the obscurity of legacy sources and the pressures of strict time constraints. To enhance the efficiency of on-call engineers (OCEs) and streamline their daily workflows, we introduced DECO -- a comprehensive framework for developing, deploying, and managing enterprise-grade chatbots tailored to improve productivity in engineering routines. This paper details the design and implementation of the DECO framework, emphasizing its innovative NL2SearchQuery functionality and a hierarchical planner. These features support efficient and customized retrieval-augmented-generation (RAG) algorithms that not only extract relevant information from diverse sources but also select the most pertinent toolkits in response to user queries. This enables the addressing of complex technical questions and provides seamless, automated access to internal resources. Additionally, DECO incorporates a robust mechanism for converting unstructured incident logs into user-friendly, structured guides, effectively bridging the documentation gap. Feedback from users underscores DECO's pivotal role in simplifying complex engineering tasks, accelerating incident resolution, and bolstering organizational productivity. Since its launch in September 2023, DECO has demonstrated its effectiveness through extensive engagement, with tens of thousands of interactions from hundreds of active users across multiple organizations within the company.

Active Learning (AL) is a user-interactive approach aimed at reducing annotation costs by selecting the most crucial examples to label. Although AL has been extensively studied for image classification tasks, the specific scenario of interactive image retrieval has received relatively little attention. This scenario presents unique characteristics, including an open-set and class-imbalanced binary classification, starting with very few labeled samples. We introduce a novel batch-mode Active Learning framework named GAL (Greedy Active Learning) that better copes with this application. It incorporates a new acquisition function for sample selection that measures the impact of each unlabeled sample on the classifier. We further embed this strategy in a greedy selection approach, better exploiting the samples within each batch. We evaluate our framework with both linear (SVM) and non-linear MLP/Gaussian Process classifiers. For the Gaussian Process case, we show a theoretical guarantee on the greedy approximation. Finally, we assess our performance for the interactive content-based image retrieval task on several benchmarks and demonstrate its superiority over existing approaches and common baselines. Code is available at https://github.com/barleah/GreedyAL., Comment: Accepted to Transactions on Machine Learning Research (TMLR)

Challenging neutron-capture cross-section measurements of small cross sections and samples with a very limited number of atoms require high-flux time-of-flight facilities. In turn, such facilities need innovative detection setups that are fast, have low sensitivity to neutrons, can quickly recover from the so-called $\gamma$-flash, and offer the highest possible detection sensitivity. In this paper, we present several steps toward such advanced systems. Specifically, we describe the performance of a high-sensitivity experimental setup at CERN n\_TOF EAR2. It consists of nine sTED detector modules in a compact cylindrical configuration, two conventional used large-volume C$_{6}$D$_{6}$ detectors, and one LaCl$_{3}$(Ce) detector. The performance of these detection systems is compared using $^{93}$Nb($n$,$\gamma$) data. We also developed a detailed \textsc{Geant4} Monte Carlo model of the experimental EAR2 setup, which allows for a better understanding of the detector features, including their efficiency determination. This Monte Carlo model has been used for further optimization, thus leading to a new conceptual design of a $\gamma$ detector array, STAR, based on a deuterated-stilbene crystal array. Finally, the suitability of deuterated-stilbene crystals for the future STAR array is investigaged experimentally utilizing a small stilbene-d12 prototype. The results suggest a similar or superior performance of STAR with respect to other setups based on liquid-scintillators, and allow for additional features such as neutron-gamma discrimination and a higher level of customization capability.

Given a simple biconnected planar cubic graph, we associate each its vertex among $2n$ ones with the so-called spin, i.e., a variable which takes on values $\pm 1$. P. J. Heawood has proved that a Tait coloring, accurate to the choice of a color for one edge, is equivalent to the choice of spin values so as to make the sum of these value at vertices of any face be a multiple of~3. We treat faces, which satisfy this condition, as {\it proper}. The condition that guarantee the propriety of faces define a system of linear equations (SLE) with respect to variables, which take on nonzero values in the field ${\mathbb F}_3$. We say that a set of vertices is {\it defining} if values of spins of these vertices uniquely define values of the rest spins. In particular, so is the set of vertices which correspond to all free variables of the SLE. We actualize the approach proposed by P. J. Heawood by proposing a geometric proof of the fact that for a non-bipartite graph the rank of the SLE equals $n+1$. Moreover, we also geometrically describe the necessary condition for the minimality of the defining set. This implies that in the case of a non-bipartite graph there exist defining subsets consisting of $n-1$ vertices. As a simple corollary, we conclude that the number of Tait colorings in this case does not exceed $3\cdot 2^{n-1}$. Though this estimate is not exact, it is by half better than the known one. We also prove that the number of Tait colorings for a graph $CL_n$, which is bipartite for even $n$ and non-bipartite for an odd one, equals $2^n+8$ and $2^n-2$, correspondingly., Comment: 12 pages, 1 figure

This document is comprised of a collection of updated preliminary parameters for the key parts of the muon collider. The updated preliminary parameters follow on from the October 2023 Tentative Parameters Report. Particular attention has been given to regions of the facility that are believed to hold greater technical uncertainty in their design and that have a strong impact on the cost and power consumption of the facility. The data is collected from a collaborative spreadsheet and transferred to overleaf.

Formal verification using proof assistants, such as Coq, is an effective way of improving software quality, but it is expensive. Writing proofs manually requires both significant effort and expertise. Recent research has used machine learning to automatically synthesize proofs, reducing verification effort, but these tools are able to prove only a fraction of the desired software properties. We introduce Cobblestone, a new proof-synthesis approach that improves on the state of the art by taking advantage of partial progress in proof synthesis attempts. Unlike prior tools, Cobblestone can produce multiple unsuccessful proofs using a large language model (LLM), identify the working portions of those proofs, and combine them into a single, successful proof, taking advantage of internal partial progress. We evaluate Cobblestone on two benchmarks of open-source Coq projects, controlling for training data leakage in LLM datasets. Fully automatically, Cobblestone can prove 48% of the theorems, while Proverbot9001, the previous state-of-the-art, learning-based, proof-synthesis tool, can prove 17%. Cobblestone establishes a new state of the art for fully automated proof synthesis tools for Coq. We also evaluate Cobblestone in a setting where it is given external partial proof progress from oracles, serving as proxies for a human proof engineer or another tool. When the theorem is broken down into a set of subgoals and Cobblestone is given a set of relevant lemmas already proven in the project, it can prove up to 58% of the theorems. We qualitatively study the theorems Cobblestone is and is not able to prove to outline potential future research directions to further improve proof synthesis, including developing interactive, semi-automated tools. Our research shows that tools can make better use of partial progress made during proof synthesis to more effectively automate formal verification., Comment: 13 pages, 10 figures

Demand for streaming services, including satellite, continues to exhibit unprecedented growth. Internet Service Providers find themselves at the crossroads of technological advancements and rising customer expectations. To stay relevant and competitive, these ISPs must ensure their networks deliver optimal video streaming quality, a key determinant of user satisfaction. Towards this end, it is important to have accurate Quality of Experience prediction models in place. However, achieving robust performance by these models requires extensive data sets labeled by subjective opinion scores on videos impaired by diverse playback disruptions. To bridge this data gap, we introduce the LIVE-Viasat Real-World Satellite QoE Database. This database consists of 179 videos recorded from real-world streaming services affected by various authentic distortion patterns. We also conducted a comprehensive subjective study involving 54 participants, who contributed both continuous-time opinion scores and endpoint (retrospective) QoE scores. Our analysis sheds light on various determinants influencing subjective QoE, such as stall events, spatial resolutions, bitrate, and certain network parameters. We demonstrate the usefulness of this unique new resource by evaluating the efficacy of prevalent QoE-prediction models on it. We also created a new model that maps the network parameters to predicted human perception scores, which can be used by ISPs to optimize the video streaming quality of their networks. Our proposed model, which we call SatQA, is able to accurately predict QoE using only network parameters, without any access to pixel data or video-specific metadata, estimated by Spearman's Rank Order Correlation Coefficient (SROCC), Pearson Linear Correlation Coefficient (PLCC), and Root Mean Squared Error (RMSE), indicating high accuracy and reliability.

Modern cryptographic methods for implementing privacy-preserving LLMs such as Homomorphic Encryption (HE) require the LLMs to have a polynomial form. Forming such a representation is challenging because Transformers include non-polynomial components, such as Softmax and layer normalization. Previous approaches have either directly approximated pre-trained models with large-degree polynomials, which are less efficient over HE, or replaced non-polynomial components with easier-to-approximate primitives before training, e.g., Softmax with pointwise attention. The latter approach might introduce scalability challenges. We present a new HE-friendly variant of self-attention that offers a stable form for training and is easy to approximate with polynomials for secure inference. Our work introduces the first polynomial LLMs with 32 layers and over a billion parameters, exceeding the size of previous models by more than tenfold. The resulting models demonstrate reasoning and in-context learning (ICL) capabilities comparable to standard transformers of the same size, representing a breakthrough in the field. Finally, we provide a detailed latency breakdown for each computation over encrypted data, paving the way for further optimization, and explore the differences in inductive bias between transformers relying on our HE-friendly variant and standard transformers. Our code is attached as a supplement.

Sparse Autoencoders for transformer-based language models are typically defined independently per layer. In this work we analyze statistical relationships between features in adjacent layers to understand how features evolve through a forward pass. We provide a graph visualization interface for features and their most similar next-layer neighbors (https://stefanhex.com/spar-2024/feature-browser/), and build communities of related features across layers. We find that a considerable amount of features are passed through from a previous layer, some features can be expressed as quasi-boolean combinations of previous features, and some features become more specialized in later layers., Comment: Presented at the Attributing Model Behavior at Scale (ATTRIB) workshop at NeurIPS 2024

Arecibo 305-m Telescope observations between 2008 and 2018 detected a radio continuum and spectral-line outburst in the nearby galaxy, NGC 660. Excited-OH maser emission/absorption lines near 4.7 GHz, and H$_2$CO absorption at 4.83 GHz varied on time-scales of months. Simultaneously, a continuum outburst occurred in which a new compact component appeared, with a GHz-peaked spectrum and a 5-GHz flux density that rose to a peak value of about 500 mJy from 2008.0 to 2012.0. Follow-up interferometric continuum images from theVery Large Array (VLA) at 10 GHz of this new continuum component determined it to be located at the nucleus of NGC 660. Subsequent High Sensitivity Array (HSA) line and continuum VLBI observations of the NGC 660 nucleus revealed a morphology that appears to be consistent with rapidly-precessing, mildly-relativistic jets from the central black hole. While requiring detailed modeling, this strongly suggests that the outburst is due to nuclear activity. From its time-scale, the shape of the continuum light-curve, and the milliarcsec radio structure, the most likely cause of the outburst is AGN-type activity of accretion of a gas cloud onto the central black hole., Comment: 18 pages, 7 figures, accepted for publication in The Astronomical Journal

The classical Muckenhoupt's $A_p$ condition is necessary and sufficient for the boundedness of the maximal operator $M$ on $L^p(w)$ spaces. In this paper we obtain another characterization of the $A_p$ condition. As a result, we show that some strong versions of the weighted $L^p(w)$ Coifman--Fefferman and Fefferman--Stein inequalities hold if and only if $w\in A_p$. We also give new examples of Banach function spaces $X$ such that $M$ is bounded on $X$ but not bounded on the associate space $X'$., Comment: v2: typos corrected

We use a c-GAN (convolutional generative adversarial) neural network to analyze transliterated text fragments of extant, dead comprehensible, and one dead non-deciphered (Cypro-Minoan) language to establish linguistic affinities. The paper is agnostic with respect to translation and/or deciphering. However, there is hope that the proposed approach can be useful for decipherment with more sophisticated neural network techniques., Comment: First draft. Comments are welcome

Phenotyping consists in applying algorithms to identify individuals associated with a specific, potentially complex, trait or condition, typically out of a collection of Electronic Health Records (EHRs). Because a lot of the clinical information of EHRs are lying in texts, phenotyping from text takes an important role in studies that rely on the secondary use of EHRs. However, the heterogeneity and highly specialized aspect of both the content and form of clinical texts makes this task particularly tedious, and is the source of time and cost constraints in observational studies. To facilitate the development, evaluation and reproductibility of phenotyping pipelines, we developed an open-source Python library named medkit. It enables composing data processing pipelines made of easy-to-reuse software bricks, named medkit operations. In addition to the core of the library, we share the operations and pipelines we already developed and invite the phenotyping community for their reuse and enrichment. medkit is available at https://github.com/medkit-lib/medkit

While research on adolescent smoking is extensive, little attention has been given to smoking behaviors among rural middle-aged and older adults. This study examines the role of personal networks and sociodemographic factors in predicting smoking status in a rural Romanian community. Using a link-tracing sampling method, we gathered data from 76 participants out of 83 in Leresti, Arges County. Face-to-face interviews collected sociodemographic data and network information, including smoking status and relational dynamics. We applied multilevel logistic regression models to predict smoking behaviors (current smokers, former smokers, and non-smokers) based on individual characteristics and network influences. Results indicate that social networks significantly influence smoking behaviors. For current smokers, having a smoking family member greatly increased the odds of smoking (OR = 2.51, 95% CI: 1.62, 3.91, p < 0.001). Similarly, non-smoking family members increased the likelihood of being a non-smoker (OR = 1.64, 95% CI: 1.04, 2.61, p < 0.05). Women were less likely to smoke, highlighting sex differences in behavior. These findings emphasize the critical role of social networks in shaping smoking habits, advocating for targeted interventions in rural areas., Comment: 27 pages; 1 figure; 6 tables

Understanding why researchers cite each other has been a longstanding conjecture in studying scientific networks. Prior research suggests relevance, group cohesion, or honest source crediting as possible factors. However, the dual nature of cognitive and social dimensions underlying citation is often overlooked by not considering the intermediary steps leading up to a citation. For one work to be cited by another, it must first be published by a set of authors. Therefore, we investigate the reasons behind researchers' citations, explicitly examining the interplay of socio-cognitive ties through the interdependence of coauthorship and citation networks. We assess our claims in an empirical analysis by employing the Author-Oriented Relational HyperEvent Model (AuthRHEM) to study Chilean astronomers' citation and collaboration behavior between 2013 and 2015 in a joint framework. We find evidence that when deciding which work to cite, authors prefer other work with novelty and cognitive ties, such as work-to-work relations. At the same time, coherent groups are relevant because coauthors are cocited more frequently in subsequent publications.

The International Muon Collider Collaboration (IMCC) [1] was established in 2020 following the recommendations of the European Strategy for Particle Physics (ESPP) and the implementation of the European Strategy for Particle Physics-Accelerator R&D Roadmap by the Laboratory Directors Group [2], hereinafter referred to as the the European LDG roadmap. The Muon Collider Study (MuC) covers the accelerator complex, detectors and physics for a future muon collider. In 2023, European Commission support was obtained for a design study of a muon collider (MuCol) [3]. This project started on 1st March 2023, with work-packages aligned with the overall muon collider studies. In preparation of and during the 2021-22 U.S. Snowmass process, the muon collider project parameters, technical studies and physics performance studies were performed and presented in great detail. Recently, the P5 panel [4] in the U.S. recommended a muon collider R&D, proposed to join the IMCC and envisages that the U.S. should prepare to host a muon collider, calling this their "muon shot". In the past, the U.S. Muon Accelerator Programme (MAP) [5] has been instrumental in studies of concepts and technologies for a muon collider., Comment: This document summarises the International Muon Collider Collaboration (IMCC) progress and status of the Muon Collider R&D programme

There is a growing body of work on learning from human feedback to align various aspects of machine learning systems with human values and preferences. We consider the setting of fairness in content moderation, in which human feedback is used to determine how two comments -- referencing different sensitive attribute groups -- should be treated in comparison to one another. With a novel dataset collected from Prolific and MTurk, we find significant gaps in fairness preferences depending on the race, age, political stance, educational level, and LGBTQ+ identity of annotators. We also demonstrate that demographics mentioned in text have a strong influence on how users perceive individual fairness in moderation. Further, we find that differences also exist in downstream classifiers trained to predict human preferences. Finally, we observe that an ensemble, giving equal weight to classifiers trained on annotations from different demographics, performs better for different demographic intersections; compared to a single classifier that gives equal weight to each annotation., Comment: To appear in the Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, ACL 2024

We explore the new proofs and extensions of the Heisenberg Uncertainty Principle introduced by A.~Widgerson & Y.~Widgerson in [MR4229152], developed in [MR4453622] by N.C.~Dias, F.~Luef and J.N.~Prata and also in [MR4337266] by Y.~Tang. In particular we give here a proof of the Uncertainty Principle for operators in the Metaplectic group in any dimension., Comment: 48 pages, 28 references

The task of Visual Place Recognition (VPR) is to predict the location of a query image from a database of geo-tagged images. Recent studies in VPR have highlighted the significant advantage of employing pre-trained foundation models like DINOv2 for the VPR task. However, these models are often deemed inadequate for VPR without further fine-tuning on VPR-specific data. In this paper, we present an effective approach to harness the potential of a foundation model for VPR. We show that features extracted from self-attention layers can act as a powerful re-ranker for VPR, even in a zero-shot setting. Our method not only outperforms previous zero-shot approaches but also introduces results competitive with several supervised methods. We then show that a single-stage approach utilizing internal ViT layers for pooling can produce global features that achieve state-of-the-art performance, with impressive feature compactness down to 128D. Moreover, integrating our local foundation features for re-ranking further widens this performance gap. Our method also demonstrates exceptional robustness and generalization, setting new state-of-the-art performance, while handling challenging conditions such as occlusion, day-night transitions, and seasonal variations., Comment: ICLR 2025

We consider how the absence of thermalisation affects the classical Coulomb blockade regime in quantum dots. By solving the quantum kinetic equation in the experimentally accessible regime when the dot has two relevant occupation states, we calculate the current-voltage characteristics for arbitrary coupling to the leads. If the couplings are strongly asymmetric, we have found that the Coulomb staircase practically reduces to the first step independent of the charging energy when the latter is larger than the Fermi energy, while the standard thermalised results are recovered in the opposite case. If the couplings are of the same order, the absence of thermalisation has a new, striking signature - a robust additional peak in the differential conductance., Comment: 16 pages, 4 figures