Computation and Language
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Sunday 8 November
The rapid growth of scientific literature and the high heterogeneity of biological terminology and mechanistic descriptions make it difficult for researchers to integrate scattered research evidence, identify connections among biological events, and construct mechanistic networks. Here, we developed a high-throughput human-AI hybrid framework, AOP-Smart-Finder, for biological event discovery, relationship mining, and Adverse Outcome Pathway (AOP) construction and evidence mapping from large-scale literature. This framework integrates Large Language Models (LLMs), AOP-Smart retrieval-augmented generation (RAG) technology, and Python-based automation, supporting two analysis modes: one based on a single compound and the other based on a predefined set of biological events. In the di(2-ethylhexyl) phthalate (DEHP) case, the system processed 12,983 articles within 30 min and included 1,093 of them, from which it extracted 1,589 biological events (including 458 official key events [KEs] and 1,131 candidate KEs) and 2,519 key event relationships (KERs). In the amyotrophic lateral sclerosis (ALS) case, the system identified 55 stressors from 437 articles and extracted 278 biological events and 548 event relationships. Following manual curation and Weight of Evidence (WoE) evaluation, a mechanistic network comprising 16 KEs and 23 KERs was ultimately constructed. This framework can either use existing AOPs as knowledge guidance for mechanistic expansion and evidence supplementation, or discover novel event connections and construct mechanistic networks around user-specified research topics. It provides a high-throughput, traceable, human-AI hybrid methodology for complex biological mechanism elucidation and AOP discovery driven by large-scale literature.
Artificial Intelligence
Separation of Skilling Context for Standard Operating Procedure-based Agents
Standard Operating Procedures (SOPs) are kept concise for human readability, so LLM agents that generate executable skills from them lack the tacit knowledge and implementation details needed for automation. Human intervention fills these gaps, but must be repeated as SOPs evolve—a problem we call SOP-skill drift. We propose managing SOPs and Contextual Knowledge (CK) separately, together with a contract-aware evaluation that distinguishes genuine violations from CK-resolved differences. On eight Indeed-derived workflows (single LLM), it raises the functional-equivalence rate from 12% to 58%. Under three SOP-revision perturbations, previously extracted CK already covers 79.5% of newly introduced automation gaps.
Read the full abstract
Nuclear Theory
Elliptic flow of $\phi$ and $\Omega$ via quark recombination mechanism in relativistic heavy-ion collisions
Multi-strange hadrons, such as the $\phi$ meson and $\Omega$ baryon, serve as unique probes for the properties of the quark-gluon plasma in relativistic heavy-ion collisions. Using a quark recombination model with the quark phase-space information parameterized in a viscous blastwave, we show that the elliptic flow $v_2$ of $\phi$ and $\Omega$ can be describled by quark recombination in a wide energy range from $\sqrt{s_{NN}}$ = 7.7 GeV to 5.02 TeV. We check the scaling properties of $v_2$ of $\phi$ and $\Omega$ and find the number-of-constituent-quark (NCQ) scaled $v_2$ of $\phi$ and $\Omega$ are approximately equal. We further present the $v_2$ of strange quarks at hadronization extracted from $\phi$ and $\Omega$ data. We find the $v_2$ of strange quarks is close to the NCQ-scaled $v_2$ of $\phi$ and $\Omega$ at $\sqrt{s_{NN}}$ = 7.7-200 GeV, but significant larger than the latter at $\sqrt{s_{NN}}$ = 2.76-5.02 TeV, suggesting that the $v_2$ of strange quarks cannot be simply equated to the NCQ-scaled $v_2$ of $\phi$ and $\Omega$. Our results demonstrate that the quark recombination mechanism is valid for $\phi$ and $\Omega$ production in heavy-ion collisions at $\sqrt{s_{NN}}$ = 7.7 GeV to 5.02 TeV, indicating the existence of constituent quark degree of freedom at these energies.
Read the full abstract
Information Retrieval
IntentLens: Grounding Underspecified Multimodal Queries for Recommendation via Tool-Augmented Reasoning
As intelligent personal assistants are increasingly deployed in real-world scenarios, recommendation paradigms are shifting from passive preference matching to inferring user intent from underspecified multimodal inputs. In practical interactions, users typically issue a reference image paired with a vague natural language query, keeping key visual attributes, personalized preferences, and item-side constraints implicit. Existing recommendation methods either leverage historical user interactions augmented with multimodal item features or assume fully articulated textual queries; both paradigms struggle to resolve underspecified multimodal queries for accurate item matching. To address this limitation, we propose IntentLens, a tool-augmented recommendation framework that grounds ambiguous multimodal queries into query-relevant evidence prior to user-item matching. IntentLens follows a ground-then-rank pipeline: a shared multimodal LLM leverages visual tools, user memory modules, and item attribute retrieval tools to (i) infer latent user intent from image-text inputs and (ii) augment each candidate item with fine-grained, query-conditioned evidence. Subsequently, a lightweight ranker scores candidates based on the grounded representations. To facilitate research in this novel setting, we construct two benchmarks with underspecified multimodal queries, namely GoogleReview-MMTool and Yelp-MMTool. Extensive experiments on these benchmarks demonstrate that IntentLens outperforms competitive MLLM and retrieval baselines, and exhibits strong generalization to unseen categories and datasets.
Read the full abstract
Combinatorics
Extremal graphs for Alon's graph coupon-collector conjecture
Let \(G\) be a finite simple graph without isolated vertices. Starting with every vertex uncovered, at each step sample a vertex uniformly and independently and cover one of its uncovered neighbors, chosen uniformly, if any. Let \(\cc(G)\) denote the expected number of steps required to cover all vertices. Alon conjectured that disjoint unions of \(K_{d+1}\) maximize \(\cc(G)\) among \(d\)-regular graphs of a prescribed order whenever such unions exist. In this paper, we disprove this conjecture for every fixed \(d\ge2\) and give a complete characterization of the extremal graphs for \(d=2\) at every order \(n\ge3\). In particular, all but at most six vertices of any such extremal graph lie in four-cycle components.
Read the full abstract
Machine Learning
Spatial Harness: Grounding Spatial Reasoning in Executable World States with Physics Engines
Spatial reasoning requires understanding how objects and observers are arranged in a scene and how their relations change with viewpoint or movement. Vision-language models (VLMs) typically perform both steps through model-internal reasoning, forming an implicit scene interpretation from partial views and imagining the consequences of spatial transformations. A misread scene can lead to a wrong answer, while even a correct interpretation does not ensure that imagined transformations satisfy spatial constraints. We introduce Spatial Harness, which represents scenes as explicit and executable world states to ground spatial reasoning in physics-engine feedback. A Perceiver integrates scene constraints extracted from the question with specialized visual estimates to place objects and cameras in a shared coordinate frame. The resulting state is instantiated in a physics engine, where a Reasoner executes viewpoint changes and uses computed measurements and rendered views to guide reasoning. We evaluate on MindCube, SAT, MMSI-Bench, and SpatialRGPT-Bench with open- and closed-weight VLMs. On MindCube, Spatial Harness improves Qwen3.5-27B from 52.70% to 71.14%, surpassing pySpatial by 7.52 percentage points. Given identical annotated object and camera poses, it reaches 98.25% versus 71.81% for model-internal reasoning. Code is publicly available at: https://github.com/tmlr-group/Spatial-Harness
Read the full abstract
Statistical Mechanics
Ergotropy as distance from equilibrium
We show that the ergotropy (a.k.a available energy) of a thermally isolated system is the $1$-Wasserstein distance between two energy probability density functions. Specifically, between the energy profile of the system state, and the energy profile of its associated passive state. We prove this for both classical and quantum systems, and illustrate it with examples. Our result formalises the intuition that ergotropy is a quantifier of distance from equilibrium
Read the full abstract
Machine Learning
Agent Self-Improvement with Frozen Weights: The Added Value and Limits of Failure Traces
We evaluate whether an AI agent's harness, the notes around its model, can learn from the agent's failure traces beyond a matched search that never sees them. It can, but less than it looks. We tested an open 9-billion-parameter model with fixed weights; a larger teacher model wrote its notes. On generated customer-service decisions, where both note writers were told the target rules, notes based on the agent's mistakes improved accuracy by 4.0 percentage points over notes from the same process with the mistakes withheld. That process alone, trying and keeping better-scoring notes, reached about 60% of the total improvement. In full customer-service conversations the same notes showed no demonstrated improvement. Separately, a strict screen of single lessons admitted two customer-service lessons, still awaiting tests on new data, and one tool-use lesson, written without seeing the agent's mistakes, which showed no demonstrated gain on new tool-use tasks. A neutral note with no procedure scored higher than the learned lessons (exploratory). A blind audit judged 146 of 240 tool-use test items defective.
Read the full abstract
Number Theory
On quantitative aspects of ideal membership certificate and GCD inequality for polynomials evaluated at almost units
In this paper, we obtain an effective bounds for the ideal membership certificate in polynomial rings over almost all rings of S-integers, thereby generalizing results of Aschenbrenner and Evertse and Gyory. We also establish an effective refinement of a result of Xiao, building on earlier work of Levin and Corvaja and Zannier. In particular, we provide explicit bounds for both the number and the total degrees of the exceptional hypersurfaces arising in an inequality that bounds the greatest common divisor of multivariable polynomials evaluated at almost S-units under consideration. As an application of our results, we obtain an effective refinement of a theorem of Corvaja and Zannier: Let $(F_n)_n$ and $(G_n)_n$ be two linear recurrence sequences and $\mathcal{N}$ be the set of positive integers $n$ such that $F_n/G_n$ is an integer. Hadamard quotient theorem, conjectured by Pisot and proven by Van der Poorten states that if the set $\mathcal{N}$ coincides with $\mathbb{N}$ then $\left(F_n/G_n\right)_n$ is itself linear recurrence sequence. Corvaja-Zannier proved that if $\mathcal{N}$ is infinite set, then there exists a nonzero polynomial $P(X)$ such that $\left( P(n)F_n/G_n\right)_n$ agrees with a linear recurrence sequence for all $n$ in a suitable arithmetic progression. We give an explicit bound $C$, depending explicitly on the parameters of the recurrence sequences such that, the conclusion holds if $\mathcal{N}$ contains at least $C$ elements.
Read the full abstract
Cryptography and Security
Agentic Attribution of Cloud Resource Ownership
Security tools raise findings across cloud environments, and AI agents now investigate and prioritize them. A finding is still resolved by people: it has to reach the people responsible for fixing it and accountable for the resource it concerns. Ownership is the organizational context that decides who that is. Where recorded, it is declared by hand in configuration management databases or owner tags and goes stale unless kept current; accounts are shared, and automation creates much of the infrastructure. The knowledge sits in work records, code history, audit logs and identity directories, encoded in each organization's own conventions. Recovering it poses four challenges: the signal is scattered across sources, none of them authoritative; the conventions that carry it differ between organizations; reaching an owner takes multi-hop links; and those related to a resource differ in the strength of their relation. We define ownership as a set of entities, people or teams, each related to the resource by one of five patterns ordered by strength. We present a reference architecture for agentic ownership attribution that derives ownership from these records instead of declaring it: it profiles an organization offline into tested signal procedures and an ownership playbook, then attributes each resource with a coordinator agent and parallel investigators, with confidence computed by rules. On 100 resources of 50 Amazon Web Services (AWS) resource types in one organization, we pool the architecture's candidates with those of an unprimed general-purpose agent and verify every evidence piece against the records. The architecture reaches 92.1% weighted recall and 86.6% weighted precision, against 54.8% and 77.4%, at 0.55 times the agent's cost per unit of coverage. The difference lies in people who build and change resources without an explicit owner, found through work records and code history.
Read the full abstract
Superconductivity
Visualizing d-wave pairing symmetry at the atomic scale in a cuprate high-Tc superconductor by Josephson tunneling
Superconducting pairing symmetry provides crucial insight into the microscopic mechanisms of Cooper pairing. However, probing the local manifestations of sign-changing superconducting order parameters remains experimentally challenging. Here, we investigate the superconducting condensate in optimally doped (OPT) Bi2Sr2Ca2Cu3O10+{\delta} (Bi-2223) using scanning Josephson tunneling microscopy (SJTM), equipped with an s-wave superconducting tip. Measurements at ultra-low temperatures reveal a pronounced modulation of the Josephson tunneling signal at different locations inside the crystallographic unit cell. A distinct Josephson tunneling signal is detected above O sites in the underlying CuO2 plane, whereas the signal nearly vanishes above the Cu-plaquette diagonals, including Cu sites and plaquette centers. This unique intra-unit-cell spatial pattern agrees with the theoretical prediction for the symmetry-dependent cancellation of Josephson tunneling between an s-wave superconducting tip and a d-wave superconducting sample. Our findings provide atomic-scale, real-space evidence for d-wave pairing symmetry in OPT high-Tc Bi-2223 and establish SJTM as a powerful approach for investigating the pairing symmetry of unconventional superconductors.
Read the full abstract
Differential Geometry
Gamma classes, central connection matrices, and tt* geometry
We review the gamma class of complex projective space, but from the point of view of topological-antitopological fusion. We deconstruct the central connection matrix which appears in the work of Dubrovin, Galkin-Golyshev-Iritani, and Cotti-Dubrovin-Guzzetti.
Machine Learning
OPSRL-SSP: Optimistic Posterior Sampling for Stochastic Shortest Path with Minimax-Optimal Regret
We study online reinforcement learning for the stochastic shortest path (SSP) problem with unknown transition dynamics. Although existing posterior-sampling methods perform well empirically on SSPs, their regret guarantees do not achieve the minimax-optimal order. We address this limitation by proposing OPSRL-SSP, an optimistic posterior-sampling algorithm for the SSP setting. The algorithm operates in epochs and uses a logarithmic, time-dependent schedule for the number of posterior samples. To induce optimism, we augment the transition model with a pseudo-state assigned an optimistic boundary value. We establish a high-probability regret bound of $\tilde{O}(B_\star \sqrt{SAK} + B_\star S^2 A)$, where $B_\star$ is an upper bound on the expected cost of the optimal policy, $S$ and $A$ are the numbers of states and actions, respectively, and $K$ is the number of episodes. In addition, we extend the analysis to general zero-cost SSPs using a universal cost-perturbation argument. The leading term matches the minimax lower bound $\Omega(B_\star \sqrt{SAK})$, thereby addressing the open problem raised by Jafarnia-Jahromi et al. [19] for posterior sampling in the SSP setting.
Read the full abstract
Machine Learning
Where Should a Gate Look? Input vs. State Gating in Diagonal State Space Models
Modern state-space models owe much of their computational efficiency to their linear, often diagonal, dynamics. Classical recurrent neural networks instead allow nonlinear state-to-state interactions, which can be beneficial when memory updates need to depend on the past context. In this paper, we ask whether state feedback through a simple gate regulating memory updates can enhance the expressivity of diagonal state-space models. To investigate this question, we compare gates that depend on the current input, the recurrent state, or both. We establish a strict expressivity separation: for scalar models, state gating solves induction-head tasks for arbitrary vocabulary size and sequence length, whereas input gating fails on a finite instance under standard stability assumptions on the state transition matrix. Experiments show that state gating retains its advantage on induction-head tasks in higher-dimensional models, whereas input gating performs better on selective copying tasks. Together, our results show that even a simple, diagonal form of state-feedback gating can unlock capabilities unavailable to input-controlled models, while input and state gating offer complementary strengths and achieve the strongest overall performance when combined.
Read the full abstract
Atmospheric and Oceanic Physics
Directional asymmetry of atmospheric pole-to-pole teleconnection
Paleoclimate proxies show that abrupt Greenland changes were followed by Antarctic responses within decades, suggesting a fast atmospheric pole-to-pole teleconnection involving shifts in the Intertropical Convergence Zone (ITCZ) and reorganization of extratropical jets. Here, we examine how the mean ITCZ position modulates this interhemispheric teleconnection. Using an aquaplanet slab-ocean model, we apply northern high-latitude forcing in climatologies with different mean ITCZ positions. The southern high-latitude response is about twice as large when the mean ITCZ is in the Southern Hemisphere. The teleconnection is therefore directionally asymmetric. The asymmetry arises in the subtropical jet response per degree of ITCZ shift, because, even with a comparable Hadley cell response, the angular momentum carried into the upper branch depends on the distance of the ITCZ from the equator, so that an ITCZ shift toward or away from the equator reinforces or offsets the jet response. Our results suggest that the southward-displaced glacial ITCZ made Arctic-to-Antarctic transmission more efficient than today, whereas the present-day ITCZ favors the reverse.
Read the full abstract
Computer Vision
MoT3DVG: A Benchmark for Outdoor 3D Visual Grounding with Motion-Aware Descriptions and Temporal Cues
3D visual grounding (3DVG) localizes language-referred objects in 3D scenes, with broad applicability in both indoor and outdoor environments. However, existing 3DVG research focuses on static indoor objects, limiting its use in dynamic outdoor autonomous driving. Motion descriptions, which capture the temporal dynamics of moving objects, offer a promising way to address this challenge. We therefore introduce MoT3DVG, a large-scale dataset for dynamic-aware outdoor 3DVG with temporally evolving motion descriptions. It contains 850 scenarios with 31,128 frames from nuScenes dataset, and provides 144,568 language prompts with motion-aware descriptions for dynamic objects across time. We further propose DynaVG, a novel framework that effectively leverages temporal cues for outdoor 3DVG. In addition to the standard modality-specific encoders, DynaVG introduces a Short-Long Integrated Dynamic Encoding (SLIDE) module before the language-point cloud alignment. SLIDE models both local motion cues and global temporal context via short-term window encoding and long-term window shifting. Extensive experiments demonstrate its state-of-the-art performance on MoT3DVG, validating the effectiveness of motion-aware encoding. Our work reveals open challenges and promising directions for future research in outdoor 3DVG.
Read the full abstract
Quantum Physics
Unbounded advantage of adaptivity in entanglement-assisted communication
Shared entanglement can reduce the amount of classical communication needed to perform distributed computation. This can be achieved either by first measuring the entanglement and then communicating classically, or by adapting the quantum measurement to the received message. Here, we introduce tasks that adaptive protocols can solve using just three entanglement-assisted messages, while the communication cost grows without bound when the quantum receiver is not adaptive. Consequently, no bounded amount of classical communication can reproduce all correlations generated by an entanglement-assisted ternary classical channel. We establish two complementary limitations on the relation between adaptive and non-adaptive quantum protocols. First, every adaptive protocol using a finite entanglement dimension admits a non-adaptive simulation at finite communication cost. Second, every adaptive protocol sending $d$ messages can be simulated to within error $\varepsilon$ using $O(d\log(1/\varepsilon))$ non-adaptive messages. Our results reveal the fundamentally unbounded advantage of adaptivity in quantum communication, but also that observing very large advantages requires both high-dimensional entanglement and very precise observations.
Read the full abstract
Optimization and Control
CVaR Deep Optimal Stopping
Risk-sensitive optimal stopping is useful when poor stopping outcomes carry substantial costs, as in option exercise, sequential medical treatment, and safety-critical control. A natural objective is Conditional Value at Risk (CVaR) of the stopped reward. We develop a theoretical framework and algorithm for CVaR deep optimal stopping. The framework and algorithm address four difficulties in applying risk-sensitive reinforcement learning to optimal stopping: state augmentation, score-function pathology, blindness to success, and dependence on auxiliary networks. We isolate three structural properties of optimal stopping: single-point reward, action-independent dynamics, and sum-zero gradient. We exploit these properties to establish Markov optimality without state augmentation and derive a pathwise gradient estimator. Under bounded policy derivatives, the pathwise estimator has bounded second moments near deterministic policies, avoiding the divergence that can occur with score-function estimators. Building on this framework, we propose DIOS (Distribution-Informed Optimal Stopping) to address the remaining two difficulties. A per-time-step soft penalty combined with tail-level annealing mitigates blindness to success. Estimating the reward quantile directly from each batch removes the need for an auxiliary critic. For a fixed tail level, we establish a finite-time stationarity guarantee under regularity assumptions along the training iterates. Across the 24 main Bermudan max-call configurations, DIOS achieves the highest mean CVaR and the shortest recorded training time among the tested CVaR baselines.
Read the full abstract
Differential Geometry
Harmonic functions and comparsion geometry for noncompact 3-manifolds with nonnegative scalar curvature
In this paper, we establish several sharp comparsion results for noncompact 3-manifolds with nonnegative scalar curvature and with a compact, connected boundary under a boundary mean curvature condition. Then we apply these gradient estimates to establish a Cohn-Vossen type bound for the average of the scalar curvature. This generalizes the work of Colding-Minicozzi \cite{Cold4} and Oronzio \cite{Oro2}.
Read the full abstract
Machine Learning
When Credit Turns Against Tools: Transition-Aware Credit Assignment in Agentic Reinforcement Learning
Agentic reinforcement learning with verifiable rewards trains language models through multi-turn tool interactions, yet their reasoning performance and tool-calling ability can decline during training. We trace this degradation to three coupled mechanisms: (1) tool-token imbalance produces negative overall tool credit when incorrect trajectories contain more tool tokens per turn than correct ones; (2) aggregation amplification can further strengthen this negative signal as token-mean aggregation gives longer incorrect trajectories greater weight in policy updates; and (3) failure-driven retries prolong trajectories as models tend to call tools again after execution errors, extending potential negative credit to recovery attempts. To counter this suppression, we propose Transition-Aware Credit Assignment (TACA), which targets the transition from reasoning to tool use through supervision of both call initiation and generation. Specifically, TACA uses Advantage Projection to enforce a nonnegative mean advantage on tool-call tokens within each prompt group through a minimal uniform shift. Additionally, TACA guides tool-use decisions with Decision Contrast, comparing rewards for tool-assisted and direct answers to the same prompt. Through Supervised Replay, it reinforces tool-call generation using calls from previously sampled correct trajectories. Across models and tasks, TACA improves performance and sustains more reliable tool interaction during training, raising mean accuracy over MT-GRPO on six reasoning benchmarks by 4.61 percentage points for Qwen2.5-1.5B and 3.61 for Qwen2.5-3B. Our code is available at https://github.com/tmlr-group/TACA.
Read the full abstract
General Mathematics
Approximation by subsequences of matrix transform means of Vilenkin–Fourier series
High Energy Physics (Experiment)
Observation of a strange $J^P=1^{-}$ resonance $K^{*}_1(1870)$ in $\chi_{c1}\to K^+K^-\eta^{\prime}$ decays
A partial wave analysis of the process $\psi(2S)\to \gamma\chi_{c1}$, $\chi_{c1}\to K^{+}K^{-}\eta^{\prime}$, with the $\eta^{\prime}$ meson reconstructed in the $\pi^+\pi^-\gamma$ and $\pi^+\pi^-\eta$ decay modes, is performed using a sample of $(2712.4 \pm 14.3)\times10^{6}$ $\psi(2S)$ events collected with the BESIII detector. In the $K\eta^{\prime}$ system, a resonant contribution with quantum numbers $J^{P}=1^{-}$ is observed for the first time and is denoted as the $K_1^*(1870)$ meson. Its statistical significance is $11.0\sigma$, and its mass and width are measured to be $M=(1869.0 \pm 14.6\pm 18.6$) MeV/$c^{2}$ and $\Gamma=(167.3 \pm 34.0\pm 24.2$) MeV, respectively. The branching fraction of the $\chi_{c1}\to K^{\mp}K_1^*(1870)^{\pm}\to K^{+}K^{-}\eta^{\prime}$ decay chain is determined to be $(1.48 \pm 0.30\pm 0.33)\times10^{-5}$. The branching fraction of the $\chi_{c1}\to K^{+}K^{-}\eta^{\prime}$ decay is measured to be $(7.90 \pm 0.48)\times10^{-4}$. Several intermediate scalar, vector, and tensor states are required to provide a good description of the data. These results provide new input for understanding strange-meson spectroscopy and investigating possible hybrid-state candidates.
Read the full abstract
Sound
DIY Customized Speech Recognition for a Person with Speech Disabilities
General-purpose state-of-the-art automatic speech recognition (ASR) systems fail systematically for speakers with moderate to severe speech disabilities. This paper describes the design, implementation, and evaluation of a customized ASR system built for the first author, a researcher with cerebral palsy and dysarthria. We fine-tuned OpenAI's Whisper medium and large-v3 models using Low-Rank Adaptation (LoRA) on audio samples recorded from the first author's own speech on consumer-grade hardware. On a held-out test set of 100 previously unseen utterances, the corresponding transcriptions resulted in a Word Error Rate drop from 49% to 17% for medium and from 30% to 12% for large-v3, while the Character Error Rate fell from 26% to 7% for medium and from 16% to 5% for large-v3, indicating that the residual errors are minor phonemic substitutions. Based on these models, the first author developed AVoz, a mobile application that acts as a real-time interpreter, converting his speech into synthesized voice for the listener. All processing is performed locally, ensuring that no audio data leaves the user's machine. We argue that the methodology is replicable. Moreover, we publish the full LoRA configuration, the implementation, and a step-by-step recording and training protocol, so that other people with speech disabilities, or their family members and caregivers, can build their own systems with a few hours of labeled audio data and a couple of days of GPU time.
Read the full abstract
Computer Vision
Conditional Diffusion-Based Proxy Representation Recovery for BCVA Prediction with Missing SLO Modality
Preoperative prediction of postoperative best-corrected visual acuity (BCVA) can support prognosis assessment and individualized decision-making in cataract surgery. Multimodal learning can integrate preoperative optical coherence tomography (OCT) images, scanning laser ophthalmoscopy (SLO) images, and clinical text to characterize patient status. However, SLO images are often missing in real-world clinical data due to acquisition or data-management issues, resulting in incomplete multimodal inputs. To address this problem, this paper proposes a conditional diffusion-based method for SLO proxy representation recovery and fusion prediction. Instead of reconstructing raw SLO images, the proposed method formulates SLO missingness as proxy representation recovery in the latent feature space. Specifically, OCT image features and clinical text features are first fused to construct an OCT-text conditional representation, which guides the reverse denoising process of the diffusion model to estimate sample-specific SLO proxy representations. The recovered SLO proxy representation is then used as a substitute for the unavailable SLO modality and incorporated into the downstream multimodal fusion prediction network for postoperative BCVA prediction. Experiments were conducted on 2,141 preoperative multimodal cataract samples, with an SLO missing rate of 58.80%. The proposed method achieved an accuracy of 0.9714 under the criterion of an absolute prediction error within 0.1 LogMAR and a mean absolute error of 0.0423, outperforming representative postoperative visual acuity prediction and missing-modality learning methods. These results indicate that representation-level proxy recovery can alleviate the impact of SLO modality missingness on postoperative BCVA prediction.
Read the full abstract
Human-Computer Interaction
Linking Event-sequence Data Claims to Analytic Provenance in Chatbots: Technique and Evaluation
Chatbots are increasingly used to analyze temporal event sequences of which website clickstreams are among the most common. Explanations are essential for understanding claims about event-sequence data. However, existing systems often require programming expertise to interpret these explanations and fail to explain the data processes underlying the data claims. We introduce a technique for linking a data claim to its analytic provenance (i.e., the analytic steps and results that produced it). It has two components: a reference table that records analytic steps and results in addressable cells, and a citation format that binds each claim to one or more cells. To operationalize this technique for clickstream analysis, we curated a set of analytic operations for clickstream analysis, organized them through a pivot-table abstraction, and developed tools that an AI agent can invoke to create a reference table and assemble a natural language response with cited data claims. To evaluate how citations affect users' assessment of data claims, we conducted two crowdsourced experiments with 90 participants each. The experiments showed that participants perceived correct data claims to be more accurate with citations when compared with no explanations, and that they perceived incorrect data claims to be less accurate with citations when compared with code and no explanations.
Read the full abstract
Machine Learning
When Sensors Fail: Coverage Auditing with Incomplete Sensor Records
Sensor outages can conceal prediction-interval coverage deficits behind acceptable marginal performance, while unavailable future readings leave some issued forecasts unscorable. We present a measurement-record audit that links issue-time outage duration, original target availability, and time-uniform statistical evidence. The audit retains complete forecast keys, batches all sensors and overlapping horizons at each measurement tick, and projects simultaneous covered-count and scored-count bounds onto a cumulative conditional expected coverage ratio. This target is distinguished from empirical coverage and outage-conditional population coverage. Selected traffic and air-quality results illustrate the gap between pooled and long-outage coverage; in METR-LA, only 7.6% of issued long-gap forecasts have originally observed targets. Of 30 primary sequential outage-state audits, 26 are inconclusive and four lack scored support. This concise presentation gives the core construction and selected aggregate evidence without the full experimental supplement.
Read the full abstract
High Energy Physics (Theory)
Stanley Deser 19th March 1931 - 21st April 2023
Stanley Deser was Ancell Professor Emeritus at Brandeis University and Visiting Professor at the California Institute of Technology. He was a world-renowned theoretical physicist whose accomplishments include the Arnowitt-Deser-Misner (ADM) formulation of general relativity, supergravity with Zumino, the dynamics of flat space with Jackiw and 't Hooft and many other contributions to the quantum theory of gravity. In the 1980s he was instrumental in securing National Science Foundation approval of the LIGO gravitational wave experiment against much opposition. He was awarded the Einstein Medal in 2015 for important contributions to general relativity, in particular the development of the ADM formalism. Other awards include: The Dannie Heineman Prize 1994; Fellow of the National Academy of Sciences 1994; Honorary Member of the Torino Academy of Sciences 2001; Member of the American Academy of Arts and Sciences 1979, Fellow of the American Institute of Physics.
Read the full abstract
Audio and Speech Processing
Capability-Guided Iterative Online Optimization of Cascaded Spoken-Dialogue Modules
Cascaded spoken dialogue systems comprise perception-and-comprehension, decision-making, and response-generation modules. We study iterative optimization of the first and last modules. Their reciprocal speech-content interfaces enable one-sided data to provide cross-module supervision: acoustic observations, such as waveforms or feature sequences, can be pseudo-labeled to train generation, whereas content-only samples can be rendered into speech to train comprehension. Because the reliable teacher-student direction varies across domains, we propose a two-stage, capability-guided pipeline for iterative online optimization. Stage 1 compares both modules through a shared intelligibility probe and obtains a domain-wise capability prior. Stage 2 activates the compatible one-sided-data route, constructs and filters supervision with the current teacher, updates the selected student, and refreshes the relevant checkpoints across cycles. We instantiate comprehension with Whisper large-v3 and generation with CosyVoice2. After only 200 steps for Whisper and 600 for CosyVoice2 of our iterative online optimization, we observe that, on the Chinese MAGICDATA-RAMC dataset, Whisper reduces CER from 14.18% to 10.77%, while CosyVoice2 reduces it from 4.78% to 4.04%. On the Japanese Common Voice dataset, Whisper reduces CER from 15.23% to 15.10%, while CosyVoice2 reduces it from 14.85% to 11.18%. These results demonstrate the feasibility of capability-guided iterative optimization of complementary spoken-dialogue modules.
Read the full abstract
Combinatorics
Monk and Pieri rules for forest polynomials
This article gives Pieri rules for forest polynomials and Monk rule for double forest polynomials. We construct binary relations $\lessdot_i$ and $\prec_i$ based on the local Tamari moves on indexed forests. These relations are explicit and constructive. For ordinary forest polynomials, the $h$- and $e$-Pieri rules are given by all chains of a fixed length along the cover relations, with certain restriction on the left or right endpoints. For double forest polynomials, our Monk rule shows every off-diagonal degree-one structure coefficient is always a single positive root, whose weight is also controlled by Tamari moves. Additionally, we introduce an insertion algorithm on positive injective words that yields a Monk rule for the corresponding slide polynomials. Regrouping this rule along Sylvester classes gives an independent Monk rule for ordinary forest polynomials.
Read the full abstract
Computer Vision
Beyond Similarity Entropy: A Rank-Guided Data-Free Quantization Framework for Vision Transformers
Vision Transformers (ViTs) have demonstrated remarkable performance, yet their computational and memory demands pose significant challenges for deployment on mobile devices. Data-free quantization (DFQ) enables the compression of ViTs without access to their original training data. Existing DFQ methods often synthesize calibration samples by maximizing the entropy of patch similarity. However, similarity entropy describes the dispersion of similarity values without ensuring that the generated samples cover diverse feature directions, potentially limiting their effectiveness for quantization calibration. To address this limitation, we propose RankQ-ViT, a rank-guided DFQ framework for ViTs. We first observe that patch similarity matrices derived from Gaussian noise tend to have lower ranks than those derived from real images. Motivated by this observation, we introduce a block-aware self-contrastive objective that promotes feature diversity during synthetic data generation without explicitly computing matrix rank at each optimization step. We then select calibration samples based on their rank and semantic consistency and use them for block-wise reconstruction of the quantized model. Experiments on ImageNet, COCO, ModelNet40, and ShapeNet demonstrate the effectiveness of RankQ-ViT across multiple architectures and tasks, particularly under low-bit quantization. The code is available at https://github.com/Runqing-forMost/rankq-vit.git