Peer Revenue Past citations are no guarantee of future returns

Papers

Fresh from arXiv, typeset properly. Highlighted words trade on the market.

Computer Science

cs Watch Computer Science (sign in to use)

Tuesday 11 August

Robotics

From Self-Normal-Positioning to Omni-Directional Tracking: Real-Time Surface Modeling Enabled Probe Tilt Control for Robotic Ultrasound Imaging

Ultrasound (US) provides real-time, radiation-free imaging, but the image quality depends strongly on how the probe is oriented against the patient body. Robotic US can reduce operator workload and improve acquisition consistency; however, most existing systems focus on normal positioning, where the probe is maintained perpendicular to the local surface. This constraint is inadequate for examinations like echocardiography, where obtaining a diagnostic view requires a non-normal probe angle. Consequently, a clinically useful robotic system must sense the local surface in real-time and preserve the desired probe orientation. Here, we propose an omni-directional probe-orientation control framework that integrates RGB-D perception, local-surface modeling, and task-space orientation control. The surface model fuses multi-view point clouds and provides a quadratic estimate of the local surface. A desired imaging direction is then encoded relative to the normal, enabling the probe to track arbitrary angles. The framework was evaluated through flat-surface tracking, phantom target-angle recovery, and in-vivo tracking of an expert selected view. Results show that the mean angular tracking error was 1.06 +- 0.66 deg. The system recovered a non-normal tilt angle of up to 44.39 +- 2.59 deg relative to the surface normal, and acquired the desired heart chamber view in the phantom and in-vivo experiments.
Read the full abstract

Computation and Language

Measure, Don't Optimize: Forecasting Recovery in LLM Unlearning

Prior white-box studies show that large language models can retain latent traces of target knowledge after unlearning, even when the knowledge is no longer expressed in their outputs. However, existing audits remain limited to one-off diagnostics: it is unclear whether these residual signals can predict future recovery under continued training or serve as reliable optimization targets. Resolving this gap is essential to determine whether internal auditing can move beyond post-hoc evaluation toward proactive risk monitoring and safer unlearning. We propose J-Access, an inference-time audit that uses the Jacobian lens to map intermediate representations into vocabulary space and measures how often target concepts remain accessible along the model's output pathway. We hypothesize that residual accessibility reflects recovery susceptibility: knowledge that remains closer to the output pathway requires less fine-tuning to restore, leading to faster recovery. We audit 398 public unlearned models spanning eight unlearning methods. We find that: (1) most unlearned models retain access above the retain-only gold level; (2) pre-attack accessibility predicts recovery speed and extent at the model level, but cannot identify which specific facts will be recovered; and (3) directly minimizing J-Access does not promote genuine deletion. Instead, the model learns to hide knowledge from the audit, producing lower audit scores but greater post-attack recovery. These findings position J-Access as a model-level diagnostic for assessing residual susceptibility in unlearned models. We argue internal audits should serve as an independent diagnostic dimension in unlearning evaluation, and should not be converted into optimization targets without validation.
Read the full abstract

All authors Zirui Song, Huaxing Liu, Xiang Wang, Shuai Li, Xinye Li, Lang Gao, Jinghui Zhang, Zheng Lu, Fengxian Ji, Xiaojun Chang, Xiuying Chen

Robotics

Top-down Traffic Scenario Generation via Joint Initial-Goal Diffusion and Trajectory Infilling

Robust traffic simulators are crucial for developing and testing autonomous vehicles to reduce the costly, labor-intensive real-world data collection process and the need for physical presence on the road. However, existing simulators require agents' initial states to generate trajectories, which limits scalability and diversity due to restrictions on the given initial states. While data-driven agent initialization has been widely studied, the generated initial states are not interpretable in terms of why the agents are initialized at those specific locations. Given known initial states, trajectory generation is also a challenging problem, as the model must learn the variability of the destination and how agents should reach it over time. In this paper, we propose TrafficDiffuser, a top-down traffic scenario generation framework that generates high-level traffic scenarios, defined by initial and goal state pairs, by jointly modeling them. The high-level scenario generation makes initial states better interpretable and reduces trajectory generation into as simple as an infilling problem. We demonstrate how the generated high-level traffic scenarios can be used, including constraining based on different trajectory modes and integrating them with existing trajectory generation models. We conduct extensive experiments on the Argoverse 2 motion prediction dataset to evaluate how well the generated outputs capture real-world distributions. In addition to generating goal states, TrafficDiffuser outperforms the next-best approach for agent initialization, reducing speed distribution distance by 55.3% and the off-road rate by 2.8%.
Read the full abstract

Artificial Intelligence

When Self-Consistency Backfires: Majority Vote Hurts the Majority of Hard Science Problems for Small LLMs

Majority-vote self-consistency (SC) lowers per-problem accuracy on most GPQA Diamond problems for small instruction-tuned models: on 56.6% of problems for Qwen2.5-7B and 65.7% for Llama-3-8B. An obvious fix is a verifier-free confidence gate that decides which problems to vote on. We report that the most natural version of that fix also fails, and we separate three signal failures that v1 treated as one. The token-entropy gate fails for a measurement reason: averaged over a chain of 602 tokens, entropy mostly measures fluency. The chain average sits 0.0005 nats from the non-answer average and 0.1695 nats from the answer token. So we measure confidence at the answer token itself. On Qwen2.5-7B-Instruct-Turbo (198 problems, 64 samples each), a sample that disagrees with its problem's plurality still emits its answer at a median margin of at least 20.4 nats, and at least 75.5% of such samples exceed 10 nats. Both quantities were pre-registered and tested once on 69 problems that no exploratory analysis had seen; both passed. The unit matters. Pooled across the benchmark, the margin does separate correct from incorrect samples: the fraction above 10 nats differs by +0.0586 [+0.0160, +0.0994]. Within a problem, paired, it does not: -0.0142 [-0.0501, +0.0215]. Token log-probabilities do carry information, but not for the within-question choice that self-consistency makes. The agreement gate's failure has no mechanism we can identify, and we leave it open. The new claims rest on one model: a registered replication on a second model was sampled but could not be evaluated, and we report that instead. Separately, three reasoning models could not be evaluated on hosted serverless inference at a small budget, each for a different measured reason. All three are downloadable, so this is a limit of metered APIs, not of what can be known.
Read the full abstract

Distributed Computing

An Event-Driven Cloud-Native Wearable Analytics Framework for Real-Time Clinical Workloads

Continuous physiological monitoring using consumer-grade wearables offers a transformative opportunity for clinical care and research, yet integration remains hindered by device heterogeneity, proprietary data formats, and strict regulatory requirements. We present an event-driven, cloud-native system designed to ingest, normalize, and analyze high-frequency vital signs from wearables at scale and without vendor lock-in. The system design proposes a multi-layered microservice architecture using cluster orchestration. Data acquisition is handled via a cross-platform mobile application that leverages native health frameworks, ensuring compatibility across fragmented device ecosystems. To address interoperability, we implement an event-driven transformation pipeline using stream processing engines and specialized services to map raw measurements to the FHIR standard for medical interoperability. Our novel dependency-aware FHIR minimization scheme reduces storage overhead while maintaining lossless resource reconstruction. Furthermore, the platform integrates a modular data analytics and machine learning layer based on a medallion lakehouse architecture, supporting the full machine learning lifecycle from real-time stream processing to model serving. Performance evaluation demonstrates that the ingestion pipeline sustains 50 full ingestion requests per second with median response times under 8 ms, satisfying the low-latency requirements for real-time patient monitoring. Our open-source implementation adheres to regulatory compliance standards through role-based access control and secure service-to-service communication, providing a robust foundation for deploying wearable-based monitoring in institutional healthcare settings for clinical decision support and research workloads.
Read the full abstract

All authors Elias Grünewald, Daniil Cherepko, Linus Gustafsson, Jakob Möhler, Oskar Rabe, Paul Robin Reichelt, Constantin Stahl, Lukasz Sztukiewicz, Louis Agha-Mir-Salim, Felix Balzer

Human-Computer Interaction

The Role of Variability in Human-Machine Interaction Experience

Human-machine interaction (HMI) requires control strategies that account for the nature of human motor behavior. Conventional shared-control and haptic-assistance methods typically ignore the stochastic nature of human behavior, potentially limiting both performance and human interaction experience. In this study, we designed an experimental setting and evaluated a novel human-variability-aware optimal controller. Participants performed a physically coupled haptic interaction task in three conditions: a controller mode that aims at conventionally reducing overall variability, a variability-aware controller mode designed to maintain human natural variability patterns, and a human-only control condition serving as a baseline. We analyzed behavioral variability, task performance, and human interaction experience. The results show that considering natural movement variability significantly increased perceived interaction quality in terms of usability while maintaining task performance. These findings highlight the importance of incorporating stochastic human movement characteristics into shared-control designs and demonstrate the feasibility and benefits of the proposed control strategy for human-centered control design of HMI.
Read the full abstract

All authors Sean Kille, Jan Lars Hagemann, Anne Voormann, Balint Varga, Andrea Kiesel, Sören Hohmann

Computational Geometry

A Probabilistic Interpretation of the Ball Mapper Graph

We introduce Probabilistic Ball Mapper, a formulation of Ball Mapper in which each data point is assigned a probability distribution supported only on the metric balls that contain it. This assignment defines both a partition subordinate to the Ball Mapper cover and a Markov kernel from the finite data space to the cover. We study two assignment schemes: a uniform-on-support rule and a localized radial-basis rule that incorporates distance to landmarks while preserving the underlying cover. Pushing the empirical data distribution through the kernel produces a probability distribution over vertices. Drawing twice, conditionally and independently, from each pointwise distribution produces a soft-overlap matrix. This matrix is symmetric, nonnegative, positive semidefinite, and has the vertex distribution as both marginals. It therefore provides a mass-normalized refinement of classical Ball Mapper overlap rather than another unnormalized edge count. For graphs constructed on a common cover, the vertex and overlap distributions can be compared directly. For independently fitted covers, we formulate Wasserstein and fused Gromov–Wasserstein-type discrepancies that account for vertex mass, landmark geometry when a common ambient metric is available, and intrinsic graph relations. For a fixed cover, we derive explicit perturbation bounds controlled by the sensitivity of the assignment rule, the magnitude of the data perturbation, and the data mass near cover boundaries. When the cover is recomputed, landmark motion creates an additional source of variation, for which we state a transport-based stability principle rather than an unconditional theorem. The resulting framework turns Ball Mapper into a probability-valued representation suitable for quantitative comparison while retaining its geometric interpretability and computational simplicity.
Read the full abstract

Machine Learning · Cross-listed from quant-ph

Generative Learning for Quantum Measurement Design

Extracting quantum information from a quantum state is a fundamental task of quantum computation, often requiring the estimation of many non-commuting observables under a finite measurement budget. For both near-term and early fault-tolerant settings, the measurement protocol must balance statistical efficiency against implementation resources such as circuit depth, connectivity, and entangling-gate count. Many existing strategies focus on two extremes: hardware-friendly product measurements with high sampling cost, and fully commuting measurements with deep circuits. Here we recast resource-constrained measurement design as a generative learning problem. We introduce FlowMeas, which uses a generative flow network to directly sample finite ensembles of shallow Clifford measurement circuits subject to a prescribed shot budget and hardware constraints. At zero entangling depth, FlowMeas learns qubit-wise commuting measurement schedules and already matches or improves leading product-measurement methods on nearly all molecular benchmarks. Allowing one or two entangling gate layers yields further reductions in energy estimation error of up to $27\%$ relative to the strongest state-independent product-measurement baseline. The learned policy can also be reused across related Hamiltonians, substantially accelerating retraining along a molecular potential-energy surface. We further obtain results for molecular Hamiltonians with up to 20 qubits and apply the framework to a compactly encoded 54-qubit interacting fermionic model, extending the demonstrated scale beyond prior molecular benchmarks. These results establish generative learning as a flexible and unified framework for quantum measurement design under practical resource constraints.
Read the full abstract

All authors Jun Dai, Olivier Nahman-Lévesque, Guillaume Rabusseau, Hong-Ye Hu, Cunlu Zhou

Social and Information Networks

Exploring the Social Life of Data: Finding Data You Can Trust

Artificial intelligence is changing the scale and tempo of scientific inquiry. Models can now search, integrate, and reason over data far beyond data repositories familiar to any individual researcher. Yet this expansion creates a prior problem: before a model can produce a trustworthy scientific result, it must locate data that are appropriate for the question, sufficiently reliable for the intended analysis, and accompanied by enough context to support responsible interpretation. As data becomes increasingly abundant, the challenge of finding data has been overcome by the challenge of finding data that you can trust. This paper explores how the social and empirical evidence that accumulates when data are used in research can be used, analogous to social trust networks, to determine fit for purpose and trust. Specifically, the paper explores data-usage graphs as a new layer of scientific data infrastructure. A data-usage graph connects datasets to the publications, people, institutions, topics, software, models, workflows, and other datasets through which they are produced and used. These connections reveal the social life of data: who has relied on a source, for which questions, in what combinations, with which methods, and with what observable impact. They can turn scattered traces of practice into data-usage descriptors that complement conventional metadata and support judgments of trust and fitness for purpose. The central claim is not that popularity establishes trust, but that this can be grown with appropriate contextual history. Usage evidence must therefore be combined with production quality, provenance, governance, semantic clarity, and community validation. The feasibility and value of data usage graphs is demonstrated by implementing the prototype data insights discovery service within the National Data Platform (NDP).
Read the full abstract

Software Engineering

GraphAlignCoder: Aligning Program and Proof Graphs for Code Generation

Code large language models (LLMs) can generate syntactically plausible programs that nevertheless violate hidden semantic constraints. Existing execution-feedback training methods identify whether a completed program fails, but provide limited supervision about how a correct solution should be organized. We introduce GraphAlignCoder, a training framework that transfers explicit correctness structure into code generation. GraphAlignCoder constructs an implementation graph that captures control and dependence among program regions. In parallel, a constrained Lean pipeline produces proof traces, from which we extract a formal proof-flow graph. The model first learns executable code together with graph-derived descriptions of why individual program regions are correct, and then consolidates this knowledge into code generation. GraphAlignCoder consistently outperforms the base model, code-only SFT, and CodeRL across all benchmarks. Compared with CodeRL, it increases the solved count from 38 to 50 on LiveCodeBench v6 and from 16 to 23 on BigCodeBench Hard, corresponding to relative gains of 31.6% and 43.8%, while also improving BigCodeBench Full from 359 to 363 tasks. The ablation study further shows that verification-graph injection produces the initial reasoning gain, while verification to code consolidation is essential for robust cross-benchmark transfer.
Read the full abstract

All authors Yueke Zhang, Zihan Fang, Kevin Leach, Yu Huang

Cryptography and Security

AI Guardrail Survival under Single-Cycle Agentic Self-Summarization

Long-running agents periodically compact their context, replacing the transcript with a model-generated summary. Recent work shows that dropping a standing safety constraint during compaction drives behavioral violations across many models (Governance Decay; Chen, 2026). We ask a finer question: under a single compaction cycle, how is a safety rule lost, and what does that imply for detection and evaluation? Our central finding is that a presence check is not a safety check: when compaction does not drop a rule outright, it often leaves something that looks like a rule but does not act like one. On behavioral replay, a degraded residue leads the model to perform the prohibited action far more often than an intact welded rule does (all-case gaps of +34 and +57 points under two replay models, both positive), category-level survival behaves like a residue, and even intact rules sometimes fail to fire, so an audit that checks only textual presence gives false assurance. Sharpening this, rule-form items are retained substantially more often than prominence-matched facts, which is exactly why presence-based checking feels adequate even though survival is not protection. Textual loss is regime-dependent (weld-or-drop with a single rule; degraded predicate-loss residues under a tighter budget), and we did not observe the hypothesized textual severing mode. Such loss is silent at runtime and detectable only by comparison with retained external ground truth (such as a constraint registry), which reveals textual absence but not whether a surviving rule still fires. We also document evaluation pitfalls where LLM-judge labels alone would have reversed a conclusion. All results concern a single compaction cycle.
Read the full abstract

Human-Computer Interaction

"I Don't Want My Mental Health App To Give Me Mental Health Barriers": Unpacking The Need For Digital Mental Health Tracking Services With And For The Blind Community

Digital mental health (DMH) tracking services promise continuous, personalized support for well-being, but their design often assumes sighted users. For the blind community, this assumption produces a distinct pattern of exclusion: services whose accessibility cannot be evaluated without first paying for them, community features that exclude the users they purport to support, and interfaces that leave users digitally literate but functionally blocked. We report on an explanatory sequential mixed-methods study of blind users' experiences with DMH tracking services in the United States. In the first phase, 93 legally blind adults completed a survey about their usage patterns, adoption decisions, and data-agency preferences; in the second, 10 survey respondents participated in semi-structured interviews. We analyzed closed-ended responses using descriptive statistics and the Kruskal-Wallis test, and open-ended and interview data using inductive thematic analysis, interpreting findings through Norman and Skinner's eHealth Literacy framework. Participants identified mindfulness, sleep, and goal-tracking services as their most-used categories, but also described recurring exclusion from the community-support features that other users value most. We argue that the framework's "computer literacy" dimension is insufficient on its own: many of our participants possessed the literacy but were blocked from applying it by design choices that predate the user. We contribute design recommendations for transparent pre-purchase accessibility evaluation, accessibility-native rather than retrofitted interfaces, and user-controlled data agency – recommendations intended not to accommodate blind users but to design DMH tracking services with them from the start.
Read the full abstract

Machine Learning

Mechanism Design for Generative Engines: From Exploitation toward Win-Win Outcomes

Generative engines are reshaping the web ecosystem by making citations a key mechanism for allocating attention, attribution, and downstream value. This creates a strategic tension: content providers are incentivized to optimize for model citation, while platforms must preserve answer quality and trustworthy attribution. We show that this tension can escalate into citation wars. In repeated simulations, state-of-the-art generative engine optimization (GEO) attacks adapt to conventional defenses by producing citation-seeking rewrites that degrade document quality and introduce unsupported claims. To study this problem, we formulate the supplier–platform interaction as a repeated Stackelberg game with partial monitoring. A local best-response analysis identifies when citation competition approaches an inert stationary outcome. Motivated by this finding, we propose a platform–creator mechanism called VCR based on verifiable-content rewards. Rather than only penalizing suspicious rewrites, the platform also credits rewrites that surface checkable factual substance, aligning creator incentives with answer trustworthiness. Experiments on three benchmarks show that VCR consistently achieves the largest Net defense-utility score, outperforming the strongest baseline by an average of 12.1 percentage points, and produces a win–win outcome under our empirical equivalence criterion.
Read the full abstract

Software Engineering

The Devil Is in the Interface: Evaluating How Tool Architecture Shapes Coding Agent Behavior

As large language models continue to improve, agentic systems are becoming increasingly important, and tools are a key design dimension because they determine how agents access information and take action in their environments. Prior work on agent tooling has primarily focused on expanding what agents can do, but has paid less systematic attention to how those capabilities are organized and exposed to the model. We refer to this latter design dimension as tool architecture. We study tool architecture in coding agents through controlled experiments on repository-level issue fixing, comparing six tool architectures that hold the underlying information and actions similar while varying how they are organized and exposed to the model, across three actors and a total of 11,700 trajectories. Our experiments show that, even when tools provide similar capabilities, tool architecture changes agent behavior: Compared to a basic architecture where the agent has only the bash tool, more structured low-level interfaces improve consistency across repeated attempts by up to 4.7 $\times$; natural-language search broadens repository exploration and increases access to relevant files by more than 11%; and Python CodeAct-style interfaces achieve similar task performance with 41.6% fewer steps and 56.3% lower token usage. By contrast, lightweight text-based cognitive-scaffolding tools, such as tools that let the agent record intermediate reasoning, have limited effect on actor behavior.
Read the full abstract

All authors Xiangzhe Xu, Hamidreza Saghir, Qianhui Wu, Marc-Alexandre Côté, Tong Wang, Kiran Lakkaraju, Kexin Pei, Xiangyu Zhang

Computational Complexity · Cross-listed from quant-ph

Evaluating QAOA expectation values can be as hard as counting optimal solutions

Evaluating expectation values is a critical task for variational quantum eigensolvers, parameterized quantum circuits, and many other quantum algorithms. We consider the well-studied case of the Quantum Approximate Optimization Algorithm (QAOA) for the MaxCut problem. Recent work of Wang et al. [arXiv:2511.20212] showed this task to be NP-hard in general for any QAOA depth $p\geq 2$, complementing past results showing efficiently computable formulas for $p=1$ with arbitrary problem graphs. We sharpen this dichotomy showing that for $p\geq 2$ exact or exponentially precise cost expectation value evaluation is #P-hard under deterministic polynomial-time Turing reductions. Hardness at $p\geq 2$ is shown to remain even for evaluating single pairwise correlators $\langle Z\otimes Z\rangle $, as well as for highly restricted sets of algorithm parameters. Our proof refines the NP-hardness construction of Wang et al. that recovers the maximum cut value from the largest exponent of a QAOA Laurent polynomial, utilizing a distinct and simpler construction that extracts a value proportional to the total number of maximum cuts, in addition to the optimal cut value. Thus we show that the QAOA expectation value hardness transition from $p=1$ to $p=2$ is not only from tractability to optimization hardness, but to that of counting optimal solutions. As an application we show our results imply analogous hardness results for computing gradients and Hessians of QAOA circuits.
Read the full abstract

Machine Learning

Reoptimization Algorithms for Contextual Bandits with Knapsack Constraints

We study new algorithms for Contextual Bandits with Knapsack. In these problems, there are finitely many types of customers, products, and resources. Each product is made from a fixed combination of resources, and resources have finite capacity. A decision maker must assign each arriving customer one out of a set of multiple possible products. Every assignment of a customer to a product will generate a random reward, which equals an unknown linear function of customer and product features, plus a noise term. The objective is to jointly learn the mean reward function, and to make online assignments to minimize the expected revenue loss relative to an optimal policy that knows the reward function. We propose a natural and simple extension of the Upper-Confidence-Bound (UCB) family of algorithms and apply re-optimization techniques. We show that by taking advantage of re-optimization, our algorithm achieves an average regret of $O(\frac{(\ln T)^3}{T})$ where $T$ is the horizon length. Our bound significantly reduces the $O(\frac{1}{\sqrt{T}})$ bound in the literature for closely related dynamic-pricing problems that are based on re-optimization.
Read the full abstract

Artificial Intelligence

From Numbers to Judgment: Specialist LLM Agents and Reinforcement Learning for European Listed Real Estate

We study whether the localized numerical operations and integrative judgments of financial analysis benefit from the same form of LLM specialization. Larix maps a 16-lens European listed-real-estate analysis framework to eight lens-aligned specialists; we compare a frontier LLM under monolithic versus specialist-decomposed prompting while holding the model, source evidence, task instructions, output schema, and scoring fixed. Across 19 firms spanning seven regulatory wrappers, decomposition improves the numerical-task aggregate by 15.8 percentage points but does not reliably improve, and can reduce, performance on judgment tasks, a pattern stable across four frozen-template dispatches; a single-agent control given the complete framework does not reproduce the numerical gain. Post-training Qwen3.5-9B with GRPO using task-aligned structured rewards then raises the development-split score by 12.0 points and the judgment aggregate by 14.2 points, with gains on all four sub-ceiling tasks; the gains transfer to unseen firms (+15.2 points overall; +40.4 on covenant stress) and to unseen regulatory wrappers (+4.3), with positive transfer on all three anti-memorization splits. Prompt-level decomposition thus improves modular numerical execution, whereas targeted parameter adaptation improves integrative financial judgment.
Read the full abstract

Hardware Architecture

The Fallacy of Independent Ceilings: Characterizing Coupled Load-Branch Stall Interaction

Branch mispredictions and data-cache misses are usually evaluated as separate bottlenecks: studies report perfect-branch or perfect-cache speedups as isolated upper bounds and often treat their product as the joint ceiling. In irregular workloads, however, hard-to-predict branches and cache-missing loads often occur in the same hot loops. Removing one penalty can expose the other: faster memory reaches mispredicted branches sooner, while better branch prediction leaves more long-latency loads in the out-of-order window. We call this interaction symbiotic stall latency (SSL). This paper quantifies when isolated ceilings fail using joint speedup synergy (JSS), the observed joint perfect-branch/perfect-cache speedup divided by the product of the isolated speedups. Values above one mean independent-ceiling analysis understates attainable gain. Across 53 simulated workloads, 70% show measurable coupling (JSS > 1), though many are near unity, especially in lower-pressure cases. With a conservative threshold, 40% exceed the independence product by more than 6%, and kernels with SSO > 20 show JSS from 1.23 to 3.29. We introduce symbiotic stall opportunity (SSO), a lightweight MPKI-based screen for workloads that merit full joint simulation. We map high-SSO workloads to four recurring software patterns: neighbor access, hash lookup, linked-structure traversal, and data-dependent modification. We connect SSL to reorder-buffer occupancy, squash rate, and commit starvation under isolated perfect modes. The resulting methodology is simple: use SSO to screen, JSS to validate, and report conditional branch-after-cache and cache-after-branch gains when evaluating branch predictors, prefetchers, caches, or coupled branch/memory mechanisms. Our contribution is a measurement framework showing when isolated perfect modes are adequate and when they understate joint performance headroom.
Read the full abstract

Machine Learning

Towards an approach to multivariate outlier detection for District Heating System data

In this paper, we test different methods for multivariate detection of outliers in the data of transmitted heat energy in the selected substation of local District Heating System, by also considering outside ambient temperature, namely Z-score (univariate, as a benchmark), Mahalanobis distances, Principal Component Analysis (PCA), Isolation Forest and Hotelling's T-squared test. The overall research aims at uncovering irregular plant operation, with a wider objective of identifying the opportunities for reducing the consumption of gas in central heating plants as well as the CO2 emission. The proposed approach considers specific domain circumstances, such as irrelevance of zero transmit-ted energy timepoints as indication of off-grid plant. The outcomes of the different methods are discussed with domain experts. It was concluded that PCA, Isolation Forest and Hotelling method provide relevant results. Finally, we adopt the ensemble method (selection based on the agreement of all three methods on the detected outliers) as the final approach.
Read the full abstract

All authors Rajko Turudija, Dušan Stojiljković, Milan Zdravković, Marko Ignjatović

Machine Learning

PAIR: Pairwise-Aware Inclusion Reweighting for Adaptive Rollout Allocation in RLVR

Reinforcement learning with verifiable rewards (RLVR) spends most of its compute generating groups of long reasoning trajectories. Recent allocators reduce this cost by assigning budgets to prompts, rollouts, or tokens according to a pointwise notion of difficulty or utility. We identify a statistical mismatch: the unclipped leave-one-out group-relative score gradient is not a sum of independent point contributions, but a second-order U-statistic over pairs of rollouts. Completing one rollout therefore reveals contrast with every other completed rollout, and adaptive endpoint selection changes which pair terms are observable. We introduce PAIR (Pairwise-Aware Inclusion Reweighting), which treats short rollout prefixes as vertices and pair-gradient terms as edges of a contrast graph. A prefix-only predictor estimates correctness and remaining token cost; a convex design chooses positive continuation probabilities under an expected suffix-token budget; and each edge induced by completed vertices is inverse-weighted by its logged joint inclusion probability. Under conditionally independent on-policy rollouts and an unclipped, unstandardized objective, the resulting estimator is design-unbiased for the complete candidate-pair gradient. Across compute-matched RLVR runs on Qwen3-1.7B/4B, PAIR improves average accuracy by +1.2 and +1.4 over the strongest pointwise allocator while using 51% and 52% fewer generated tokens than full-group GRPO. A frozen-population estimator audit confirms that unweighted adaptive selection is biased, whereas pair-inclusion correction recovers the complete-pair target at matched suffix cost.
Read the full abstract

All authors Pixel Nomand, Elena Voss, Marcus Hale, Sofia Reyes

Computer Vision

Gaze Target Estimation Anywhere with Concepts

Estimating human gaze targets from images in-the-wild is an important and formidable task. Existing approaches primarily employ brittle, multi-stage pipelines that require explicit inputs, like head bounding boxes and human pose, in order to identify the subject of gaze analysis. As a result, detection errors can cascade and lead to failure. Moreover, these prior works lack the flexibility of specifying the gaze analysis task via natural language prompting, an approach which has been shown to have significant benefits in convenience and scalability for other image analysis tasks. To overcome these limitations, we introduce the Promptable Gaze Target Estimation (PGE) task, a new end-to-end, concept-driven paradigm for gaze analysis. PGE conditions gaze prediction on flexible user text or visual prompts (e.g., "the boy in the red shirt" or "person in point [0.52, 0.48]") to identify a specific subject for gaze analysis. This approach integrates subject localization with gaze estimation, and eliminates the rigid dependency on intermediate analysis stages. We develop a scalable data engine to generate Gaze-Co (Gaze Estimation with Concepts), a dataset and benchmark of 120K high-quality, prompt-annotated image pairs. We also propose GazeAnywhere, the first model designed for PGE. GazeAnywhere uses a transformer-based detector to fuse features from frozen encoders and simultaneously solves subject localization, in/out-of-frame presence, and gaze target heatmap estimation. GazeAnywhere achieves state-of-the-art performance on multiple PGE benchmarks, setting a strong baseline for this new problem even on a difficult out-of-domain, real-world clinical dataset. GazeAnywhere is open-sourced in github.com/IrohXu/GazeAnywhere.
Read the full abstract

All authors Xu Cao, Houze Yang, Vipin Gunda, Zhongyi Zhou, Tianyu Xu, Adarsh Kowdle, Inki Kim, James M. Rehg

Human-Computer Interaction

QUARTZ: Qualitative Understanding via Accessible Representation and Visualization

Qualitative data visualizations – concept maps, network graphs, Sankey diagrams, and coding stripes – are integral to research practice, yet remain entirely inaccessible to blind and low-vision (BLV) researchers. While visualization has seen advanced multimodal solutions for quantitative charts, qualitative visualizations, and their non-linear, semantically rich structures have received no attention. We present QUARTZ, a web-based system that provides screen-reader-accessible, multimodal representations of qualitative data visualizations. Using the Rapid Iterative Testing and Evaluation (RITE) method, we conducted a user study with 8 BLV participants who completed 12 tasks across four visualization types. Our findings expose accessibility barriers unique to qualitative visualizations – non-linear navigation breakdowns and semantic comprehension gaps absent from quantitative chart research—and document how iterative co-design with BLV users resolved them. We contribute empirical evidence and design guidelines for an underexplored visualization domain, advancing the infrastructure BLV researchers need to participate independently in qualitative inquiry.
Read the full abstract

Robotics

What Does It Take For A Generalist Robot Policy To Adapt Using Minimal Data?

A central goal in robot learning is to move beyond task-specific human data collection toward robots that improve through autonomous interaction. Yet fully autonomous learning remains difficult with current policies: sparse rewards and weak zero-shot exploration often prevent online learning from discovering successful behavior. As a first step towards fully autonomous learning, we propose the minimal-data adaptation regime, in which a pretrained robot policy must learn a new task from as little as one demonstration followed by autonomous online interaction. We establish that, although pretrained policies fail zero-shot on many new tasks, a single demonstration suffices to direct their behavioral prior to coarse task-following behavior, albeit with unreliable performance. From this insight, we isolate the key design principles required to convert coarse task-following into reliable task success. These include (1) the surprising efficacy of pretrained VLA representations to support sample-efficient value learning from online interaction (2) the sufficiency of residual policy learning to learn corrective behavior and (3) the need for careful data-balancing to prevent exploration collapse. As a representative of these principles, we propose MiDAS, a simple offline-to-online RL recipe that combines few-demonstration behavior cloning with value-based residual learning over frozen VLA representations. Across LIBERO and RoboCasa tasks, MiDAS substantially improves performance from single-demonstration initializations, including policies that succeed only rarely, validating the principles highlighted above. On a bimanual YAM platform, starting from a brittle initial policy, MiDAS improves task success after approximately six hours of online interaction.
Read the full abstract

All authors Shreyas Kowshik, Sreyas Venkataraman, Leo Wang, Niharika Pant, Max Simchowitz, Aviral Kumar

Computational Complexity

RevCRN: Reversible Analog Computation using Chemical Reaction Networks

The computability of real numbers and functions using Turing Machines has been a central area of theoretical computer science since the mid-20th century. In the late 20th century, it was shown that chemical reactions can serve as a basis for computation using the Chemical Reaction Network (CRN) model. Recent advances in computing real numbers using Deterministic Chemical Reaction Networks (DCRNs) have identified numerous classes of DCRN-computable real numbers. In parallel, the works of R. Landauer and C. H. Bennett, spanning the 1960s to the early 2000s, showed that reversible computing offers significant advantages over irreversible methods, particularly in energy efficiency, motivating extensive research on reversible computation. In this work, we investigate the computability of real numbers using Reversible Chemical Reaction Networks (RevCRNs). The paper has two primary contributions: (1) establishing relationships among CRN-computable real number classes including Lyapunov CRN ($\mathbb{R}_{LCRN}$), Real-Time CRN ($\mathbb{R}_{RTCRN}$), rational numbers ($\mathbb{Q}$), and RevCRNs ($\mathbb{R}_{RevCRN}$), with key results: (i) $\mathbb{Q}$ is a strict subset of $\mathbb{R}_{RevCRN}$; (ii) the set of positive algebraic numbers ($ALG$), $\mathbb{R}_{LCRN}$, and real numbers computable by 1-species RevCRN ($\mathbb{R}_{RevCRN}^{1s}$) are equal; (iii) $\mathbb{R}_{RTCRN}$ and $\mathbb{R}_{RevCRN}$ exhibit non-empty overlap; and (iv) the set of real numbers computable by detailed-balanced RevCRNs ($\mathbb{R}^{DetBal}_{RevCRN}$) is a subset of $ALG$; and (2) exploring the existence of a hierarchy within $\mathbb{R}_{RevCRN}$. Finally, we leave open the exact relationship between $\mathbb{R}_{RevCRN}$ and $\mathbb{R}_{RTCRN}$ while conjecturing a general hierarchy of RevCRN-computable reals.
Read the full abstract

Machine Learning

Lifecycle-Optimal Tokenization: Vocabulary Size as a Deployment-Regime-Dependent Infrastructure Parameter

Tokenizer vocabulary size is a foundational design choice in large language model (LLM) infrastructure, yet it is typically fixed at training time based on convention rather than deployment analysis. We show that the cost-optimal vocabulary is not a constant but a function of the serving regime. We formalize total deployment cost as $C_{lifecycle}(V) = C_{train}(V) + \lambda \cdot C_{infer}(V, B)$, where $\lambda$ is inference volume and $B$ is the serving batch size. Through controlled experiments on two GPU families spanning the memory-bound to compute-bound regimes (A10G, ridge $\approx$ 117 FLOP/byte; A100, ridge $\approx$ 183 FLOP/byte), we demonstrate: (1) the inference-optimal vocabulary shifts 16x with serving batch, from 32k at $B=1$ to 524k at $B=64+$, driven by amortization of the $V \times d$ unembedding matrix read; (2) at 1.3-2.3B model scale, quality (bits per byte, BPB) is optimized at $V=65$k, confirming scale-dependent vocabulary preference; (3) the lifecycle-optimal vocabulary diverges from training-optimal by up to 16x for production deployments. Quality is approximately invariant across the optimal range ($<$2% BPB spread), making vocabulary a pure systems optimization with no quality penalty in the measured range. Our results provide actionable capacity planning guidance: on-device deployments ($B=1$) should use $V \approx 32$k; datacenter serving ($B \geq 64$, $\lambda \geq 10$) should use $V \approx 131$-262k.
Read the full abstract

Systems and Control

Time-Optimal Control of One-Mode Flexible Structures: Analytical Solution and the Cost of Flexibility

The time-optimal solution for a double integrator is a foundational result: its time-displacement law $T_r = 2\sqrt{L}$ for a normalized step $L$ is structure that numerical methods cannot provide. We solve the time-optimal control for rest-to-rest maneuvers of a double integrator coupled to a harmonic oscillator, deriving the first analytical solution and closed-form time-displacement law. Synthesis reduces to a single scalar inversion. This is the model to which the two-mass-spring system, the one-bending-mode flexible structure, and the linearized overhead crane reduce. A Pythagorean identity $T^2 = T_r^2 + 2T_s^2$ decomposes the optimal maneuver time into the rigid-body minimum and a synchronization time $T_s \leq \pi$, the cost of flexibility, giving the sharp envelope $2\sqrt{L} \leq T \leq 2\sqrt{L + \pi^2/2}$. The penalty is scale-dependent. For small maneuvers $T \propto L^{1/4}$: one oscillator mode costs as much as two additional integrators. For large ones, flexibility is asymptotically free. It vanishes at the natural motions, where the oscillator completes whole cycles. A closed-form sensitivity formula proves the maneuver time non-monotone in stiffness. Adding flexibility to a rigid body never shortens a maneuver, yet softening an already flexible structure can: stiffer is not always faster. The natural motions are robust design targets since first-order sensitivity to stiffness vanishes there.
Read the full abstract

Machine Learning

Market-Information-Aware Gated-LoRA of Foundation Models for Transferable Day-Ahead Electricity Price Forecasting

Electricity price forecasting is crucial for market participants but remains difficult because prices are volatile, market-specific, and closely tied to anticipated system conditions. Existing supervised methods depend largely on market-specific historical data, limiting their use in newly established or data-scarce markets. This paper proposes a market-information-aware adaptation framework that transfers the Chronos-2 time-series foundation model to day-ahead electricity price forecasting. It first constructs a multi-source market information (MSMI) interface aligning 7-day price context with pre-clearing supply–demand, reserve, maintenance, generator-capacity, and intertie variables, and then trains a source-domain gated low-rank adapter (LoRA), updating about $1\%$ of model parameters without target-market labels. The gate scales the frozen source adapter according to reserve-tightness and operating-state signals. A leave-one-market-out protocol is adopted for evaluating cross-market transferability. Experiments on four Chinese provincial day-ahead spot markets show that the proposed framework reduces the average MAE/RMSE by $6.24\%/7.99\%$ relative to market-information-aware zero-shot Chronos-2 and by $3.05\%/3.52\%$ relative to vanilla Source-LoRA. Experiments show that the gain is not reproduced by a learned global scalar or by random gate initialization, while the additional improvement over Source-LoRA is limited. These results suggest that market-structured inputs and state-dependent gated LoRA can provide a practical transfer path for data-scarce electricity markets.
Read the full abstract

Multiagent Systems

When Do Institutions Beat Intelligence?

More capable agents do not necessarily form a more capable collective. A multi-agent system may jointly possess sufficient information yet fail because evidence is poorly routed, unreliable reports enter public belief, correlated claims masquerade as independent support, shared state becomes stale or strategically distorted, or useful evidence is exposed through an ineffective action interface. We ask when additional resources should improve the reasoner and when they should instead change the institutional structure through which the collective forms and acts on public information. Drawing on functional distinctions from research on group decision making and distributed cognition, we construct controlled artificial ecologies around four loci of collective failure: access and routing, admission and dependence, state maintenance and incentives, and representation and action. Across these ecologies, we separately vary model capability and institutional structure, pairing positive interventions with matched reasoning baselines and mechanism-breaking controls. The experiments reveal a consistent boundary: institutions help when they repair failures in how a collective constructs usable public state, but lose their advantage when their signals are uninformative or uncheckable, when stronger intelligence can perform the same transformation directly, or when the resulting state cannot support reliable action. Our results recast the choice between intelligence and institutions as a diagnosis of where collective reasoning fails.
Read the full abstract

Artificial Intelligence

Inverse Theory of Mind Modeling for Content Recommendation: From Web Browsing to Dynamic Intelligent Interfaces

Modern recommender systems treat observed actions as reliable proxies for user preferences, yet interactions often reflect exploration or comparison rather than stable preference expression. As interfaces evolve from static layouts toward generative UIs and immersive extended reality (XR), the need for deeper, modality-agnostic user understanding grows: these adaptive environments must decide not only what to present but where, when, how prominently, and most importantly why a user acts. We propose an Inverse Theory of Mind (IToM) pipeline that reasons backward from observed interactions to infer the beliefs, preferences, and decision-making traits that explain behavior. The pipeline reconstructs each user's decision context, including what was chosen and what alternatives were available, applies LLM-driven counterfactual reasoning to produce evidence-grounded natural-language belief statements, and synthesizes these beliefs through multi-hypothesis abductive inference into a structured user persona. We evaluate on the OPeRA dataset against ground-truth personality assessments, attitudinal surveys, and interview-based personas across four tasks: next action prediction, shopping attitude alignment, Big Five personality inference, and held-out category prediction. Results show that inferred personas match or exceed ground-truth personas and that multi-hypothesis reasoning is essential for accurate personality prediction. We further demonstrate cross-modal transferability with a persona-driven spatial banking application on VisionOS.
Read the full abstract

All authors Mengyu Chen, Feiyu Lu, Chun-Fu Chen, Lucas Vinh Tran, Jay Katukuri

Numerical Analysis

Exact characterisation of maximum-angle conditions for spherical finite element meshes

Maximum-angle conditions are standard finite-element mesh hypotheses that permit anisotropic triangles excluded by minimum-angle or shape-regularity assumptions. For exact spherical triangles, however, the angles of the chordal affine core and the intrinsic spherical angles need not coincide, while radial geometry introduces curvature-scale distortion. We give an algebraic characterisation of the intrinsic spherical maximum-angle condition through a dimensionless quantity computed from the three vertex vectors. A uniform positive lower bound on this quantity is equivalent to a uniform spherical maximum-angle bound and requires neither spherical-angle nor spherical-area evaluation. We identify the support-plane geometry linking the chordal circumradius to radial distortion and derive sharp comparisons between the intrinsic spherical and chordal semi-regularity parameters. In particular, a locality-independent comparison holds with sharp constant $2/\sqrt3$ and a characterised equality case. An area-based parameter is shown to be smaller than the spherical semi-regularity parameter, with sharp constant one in the local flat limit. For spherical finite-element meshes, the vertex criterion implies uniform chordal semi-regularity with an explicit constant, while relative refinement makes the radial-distortion factors converge uniformly to one. Thus, the intrinsic criterion, together with relative refinement, provides the geometric controls used in anisotropic finite-element analysis without imposing a minimum-angle or shape-regularity condition.
Read the full abstract