Peer Revenue Publish or profit

Papers

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

All of arXiv

Friday 4 September

Computer Vision

Reachability Is Not Generalization: Understanding Verb–Noun Decomposition in Assembly Action Recognition

Assembly actions are compositional: they combine a manipulation with a part or tool. In deployment, systems routinely encounter novel combinations of familiar components, yet an atomic action classifier assigns every unseen combination exactly zero probability by construction. The prevailing solution is verb–noun decomposition, which predicts components separately and recombines them to reach unseen actions. While widely adopted, how decomposition generalizes under compositional shift remains poorly understood. We present a systematic analysis of verb–noun decomposition across three assembly datasets (MECCANO, HAViD, and IMPACT). Although decomposition escapes the atomic ceiling, its generalization extends only partially beyond it. Unseen-composition performance remains strongly tied to the co-occurrence structure of the training data, indicating that much of the observed gain arises from interpolation within densely supported regions of the compositional space rather than from unconstrained recombination. Across datasets, failures consistently concentrate on the larger-vocabulary component, and IMPACT's verb-heavy vocabulary reverses the bottleneck from nouns to verbs. We further show that shared-encoder training introduces component entanglement, encouraging reliance on co-occurrence patterns that transfer poorly to unseen compositions and trailing independent recombination by up to $6.0\times$ in harmonic mean. Taken together, these findings explain why decomposition achieves only partial compositional generalization in practice. By identifying primitive support, vocabulary asymmetry, and component entanglement as connected sources of error, we provide a portable diagnostic framework for studying compositional recognition beyond aggregate accuracy. Code: https://github.com/hisalaheiyo/assembly.
Read the full abstract

Distributed Computing

Pushing CPU Speech Synthesis to the Wall: Extreme Inference Tuning under Serverless Architecture and Billing

Instance-billed serverless platforms charge for CPU and memory over the lifetime of a warm instance, making idle inference state a direct serving cost. We present billing-aware neural text-to-speech (TTS) serving on serverless CPUs, optimizing CPU-seconds and GB-seconds rather than throughput or latency alone. Conventional runtimes are poorly suited to this setting: per-request parallelism causes CPU contention under concurrency, while warm instances retain gigabytes of billable inference and page-cache state. We address these costs with request-sized concurrent inference, which bounds per-request CPU parallelism, and a reclaimable instance lifecycle, which releases inference state and page-cache memory after idle periods while retaining the server process and compile cache. On Kokoro-82M, our system achieves 2.71 audio-seconds per CPU-second versus 0.89 with ONNX Runtime defaults and reduces cost per audio-hour from $0.0631 with PyTorch to $0.0153, a 4.1x reduction. Idle billed memory falls from 8.7 GB to 1.33 GB, while restoration reaches first audio in 2.2 s versus 7.7 s for a PyTorch cold start. Under bursty traffic, lifecycle reclamation is essential for translating inference efficiency into lower serverless cost.
Read the full abstract

Solar and Stellar Astrophysics

Asteroseismic Detection of a Massive Unseen Companion to the $\delta$ Scuti Star TIC 160582982

We present TIC 160582982, an eccentric $\delta$ Scuti binary with a dynamically massive unseen companion. Six coherent high-frequency pressure modes used as pulsation clocks yield a phase-modulation orbit with $P_{\rm orb}=89.29\pm0.19$ d, $e=0.539^{+0.040}_{-0.037}$, and $a_1\sin i/c=145.6^{+3.7}_{-3.4}$ s, corresponding to $K_1=42.2^{+2.2}_{-1.8}$ km s$^{-1}$ and $f(M)=0.416^{+0.032}_{-0.028}\,M_\odot$. An independent frequency-modulation analysis of the two strongest modes gives a consistent orbital solution. Rotating MIST isochrones constrained by the atmospheric parameters and spectral energy distribution (SED) give $M_1=1.93\pm0.19\,M_\odot$ and $R_1=2.65\pm0.23\,R_\odot$. The mass function implies a formal edge-on minimum companion mass of approximately $1.80\,M_\odot$. Assuming a luminous main-sequence companion and including the non-eclipsing geometry gives $M_2=2.19^{+1.24}_{-0.34}\,M_\odot$ and $i=61.5^{+16.6}_{-19.6}$ deg. However, no clear evidence for the presence of a luminous counterpart is found, either photometrically or spectroscopically. This suggests the presence of a massive compact object, such as a neutron star or even a stellar-mass black hole, as the massive unseen binary companion. Future phase-resolved high-resolution spectroscopy will be crucial for obtaining an independent orbital constraint and clarifying the nature of the companion.
Read the full abstract

All authors Chenglong Lv, J. Pascual-Granado, Ali Esamdin, Angela Gardini, Haozhi Wang, Jianwei Yao, Shuguo Ma

Artificial Intelligence

Gradient-Aligned Pair Selection for Personalized Preference Optimization

Personalizing large language models (LLMs) requires aligning generation behavior with user-specific preferences rather than aggregate quality. While Direct Preference Optimization (DPO) provides a stable framework for preference learning, its effectiveness in personalized settings critically depends on how preference pairs are selected. Existing approaches typically rely on heuristic criteria, such as likelihood-based extremes, which decouple optimization from explicit user utility and can lead to degraded personalization. We formalize personalized preference learning as a geometry-aligned optimization problem by analyzing the first-order interaction between gradients of expected user utility and DPO update directions. Our analysis reveals that, under off-policy sampling, the DPO update transitions from a purely error-corrective signal to a reinforcement-like update when preference margins are directionally aligned with utility gradients. This perspective exposes pair selection as a geometric decision that governs whether preference optimization advances or hinders personalization. Motivated by this insight, we propose GAP-DPO (Geometry-Aligned Preference DPO), an iterative algorithm that performs utility-aware, geometry-aligned pair selection while controlling distribution shift via epoch-wise regeneration. Experiments on personalized text generation benchmarks show that GAP-DPO consistently improves stylistic fidelity, preference alignment, and generation quality compared to standard DPO variants. Together, our results establish gradient alignment as a unifying principle for personalized preference optimization and demonstrate that pair selection is an intrinsic component of the optimization geometry rather than a heuristic preprocessing step.
Read the full abstract

All authors Ruoming Jin, Xinyu Li, Hao Zhou, Jianfeng Zhu, Ruixin Guo, Feodor Dragan, Lei Xu, Haixun Wang, Yang Zhou

Statistics Applications

Kumaraswamy-Teissier Distribution: Bayesian and Non-Bayesian Inference, Regression Modeling and Climate Application

This study introduces the three-parameter Kumaraswamy–Teissier distribution (KTD) by applying the Kumaraswamy-G construction to the Teissier distribution. The proposed distribution provides additional shape flexibility for modelling positive-valued observations and accommodates different density and hazard-rate shapes. Several statistical properties, including moments, the moment generating function, quantile function, and order statistics, are derived. Maximum likelihood, maximum product spacing, and Bayesian methods are considered for parameter estimation, and simulation results demonstrate improving estimation accuracy with increasing sample size. A median-based regression model is further developed for the KTD, with its finite-sample performance examined through simulation. The practical utility of the proposed regression model is demonstrated through an application to monthly gridded rainfall data from the Northwest Himalayas, incorporating annual and semi-annual harmonic terms to capture seasonal variation. Its performance is assessed against Kumaraswamy and beta regression using residual-based goodness-of-fit diagnostics and likelihood-based model selection. The results demonstrate effectiveness of the proposed distribution for modelling heterogeneous and skewed positive-valued rainfall data.
Read the full abstract

All authors Kamana Mishra, Neeraj Poonia, Tanmay Kayal, Sarita Azad

Systems and Control

Identification of the Steering and Speed Systems of a Four-Wheel-Steering Tractor for Optimal Control

Model-based control design requires sufficiently accurate model of the system-to-be-controlled. This paper addresses the system identification of steering and speed control of a four-wheel-steering agricultural tractor for the purpose of developing path tracking control. To this end, we investigate the steering and speed control systems of the tractor. These are complex systems with digital, mechatronic, and hydraulic components challenging to model based on first principles. We take a data-driven approach to estimating the system models. The resulting model combines the kinematic model of the vehicle, actuators modelled as first-order systems, and estimated values for time constants and transport delays.
Read the full abstract

Signal Processing

Terramechanics-Inspired Sensor Fusion for Slip and Mass Flow Estimation in Agricultural Conveyor Systems

Precision farming requires accurate knowledge of implement states. In organic fertilizer spreaders, heterogeneous material and variable field conditions complicate flow-rate prediction, and tilt angles alter the slip behaviour between transport floor and material. This paper presents a model-based sensor fusion method for slip and mass-flow estimation, integrating weighing, conveyor-speed, and inertial measurements in an Extended Kalman Filter with a volumetric flow model and a longitudinal dynamics model. Validation on real-world cattle-manure spreading shows mass estimates tracking the raw weight with an RMSE of 11.9 kg and integrated mass flow within 8.4 % of the observed mass loss.
Read the full abstract

Robotics

Multi-Reference Path Tracking Control for an Agricultural Tractor with Nonlinear Model Predictive Control

Guiding a tractor along a predefined reference path is a key component of precision agriculture. This study develops a path tracking controller based on Nonlinear Model Predictive Control, which incorporates multiple segments of a piecewise-linear reference path directly into the objective function. In addition, methods for selecting viable reference segments from the full path are presented. The control system is evaluated during a field test with a tractor controlled via the Tractor Implement Management steering interface. The NMPC solver converged on average after 3.45 ms and tracked the curved reference path with a mean absolute cross-track error of 6.1 cm.
Read the full abstract

Computational Engineering

Agentic AI for Staged Three-Dimensional Finite-Element Tunnel Modelling

Three-dimensional finite-element (FE) analysis gives the most complete picture of tunnelling-induced ground movement, but it is slow: an engineer must drive the PLAXIS 3D interface through project set-up, stratigraphy, materials, structural elements, meshing and staged calculation. This paper presents a pipeline in which a large language model (LLM) drives PLAXIS 3D through the Model Context Protocol (MCP), with no human operating the interface. A purpose-built MCP server exposes the official remote-scripting API as 19 typed tools. Four agents share the work: orchestrator, geometry, calculation and verification. They draw on 15 inspectable domain-knowledge skill modules and a provenance-tagged intake stage. The verification agent is information-barriered, auditing every model against the raw specification at a pre-mesh checkpoint and again as a calculation gate. It was evaluated on 12 single-tunnel problems in Bangkok subsoil across four capability tiers, using a 15-arm ablation matrix; 13 automated arms and a manual expert baseline were executed, one run per cell. The full system built 12 of 12 models at a mean of 6.1 min with a clean audit on every one; the same models took 24.6 min by hand. Domain skills were decisive: with the complete architecture but all 15 modules withheld, no model was produced. Base-model capability acted as a threshold rather than a gradient. Fable 5, Opus 5 and Sonnet 5 each built 12 of 12, while Haiku 4.5 halted on all 12. Among models that built, agreement was exact: 107 of 111 comparable cells reproduced the reference mesh to the element, and 83 of 86 solved cells agreed with the hand-built expert model within 1.01 % in maximum settlement and 1.21 % in trough-width parameter. Every exception traced to a value misread at intake. The reliability problem in agentic FE automation therefore lies in specification reading, not in code generation.
Read the full abstract

All authors Ochok Duangsano, Kuo-Chieh Chao, Sompote Youwai, Nattavich Sittiamornporn, Chana Phutthananon, Pornkasem Jongpradist

Computational Engineering

A deflection-basin fusion transformer for predicting critical pavement strains from falling weight deflectometer

The two critical strains of pavement design, under the asphalt layer and on the subgrade, are conventionally obtained by backcalculating layer moduli from a falling weight deflectometer (FWD) basin and propagating them through layered-elastic analysis. That route is ill-posed and confined to the office. We propose the deflection-basin fusion transformer (DBFT), a 0.22M-parameter attention model predicting both strains in one step from the basin and the layer thicknesses, read as a sequence indexed by offset. One objective spans 14,174 layered-elastic solutions and 7,651 field measurements from 96 Thai highway sections, three routes withheld entirely. On exact theory DBFT reaches R^2 >= 0.9998 at RMSE <= 1.24 microstrain, an order of magnitude better than published index equations. Trained on that factorial alone it fails on measured basins, R^2 <= 0.14: the field domain inverts the sign of the outer-basin correlation. The combined objective restores 0.964 and 0.879 on the withheld routes, holding R^2 >= 0.993 on theory. Attention and SHAP recover recognised mechanics, including reliance on the 1200-1800 mm range routine indices never reach. Agreement is with backcalculation-consistent mechanics, not gauge-measured strain. Both strains return in 1.6 ms on one CPU thread.
Read the full abstract

Computation and Language

The First Token Is Not the Verdict: Hidden Costs of Reading LLM Judges Without Generating

Reading an LLM judge's verdict from the logits of its first generated token is cheap, requires no generation, and is exactly what constrained decoding and likelihood-scoring evaluation harnesses produce. We show that this readout distorts position bias in one direction: it overstates it in every condition we test, so figures obtained this way behave as upper bounds. The mechanism is that judges do not always lead with a verdict token, on 12% to 49% of pairs for three Qwen3 judges and under 3% for Llama-3.1-8B and Phi-3.5-mini, and forcing a read on those pairs returns whichever response was shown first rather than a judgment. Pooled over the 924 pairs where a judge did not commit, the forced read flips on 89.7% of them when the responses are swapped, against 47.5% read after generation (paired difference +0.422, 95% CI [+0.365, +0.467]). The distortion is specific to what is measured: it moves position bias by 42 points while moving judge accuracy by under one point in seven of ten conditions, so it misleads whoever audits a judge rather than whoever uses one. A second, smaller failure occurs even when the judge does lead with a verdict token, since it sometimes opens with one letter and reasons its way to the other, on 0 to 5.5% of pairs at a rate uncorrelated with compliance. We recommend reporting the rate at which a judge leads with a verdict token, which costs one forward pass and no labels, alongside any position-bias figure.
Read the full abstract

Machine Learning

Format-Aware Fusion for Fast FP4 Pretraining

Four-bit floating-point (FP4) Tensor Cores accelerate matrix multiplication, but scale computation, operand packing, layout construction, and saved backward state can erase the gain. We present format-aware fusion, which co-designs each quantization producer with its scale domain and consumer layout for native \mxfp{}, global \nvfp{}, and cooperative-thread-array-local \nvfp{}. We evaluate Llama-3-family 8B pretraining through 160 billion tokens using bfloat16 output projections and compiled cross entropy. In matched same-accelerator probes, bfloat16 and Transformer Engine \nvfp{} reach 18.8K and 27.6K tokens/s/GPU, while our fastest custom route reaches 37.9K. \mxfp{} with row-gradient stochastic rounding and fixed-sign 32-value Hadamard weight-gradient preconditioning reaches 37.2K tokens/s/GPU (86.3% bfloat16 model FLOP utilization) and ends 2.11% above the raw bfloat16 training-loss endpoint. A Transformer Engine recipe with four final bfloat16 blocks ends 0.87% above bfloat16 at 27.1K tokens/s/GPU. Downstream rankings differ from training-loss rankings, showing that FP4 outcomes depend jointly on scale contract, operand, and execution path.
Read the full abstract

Information Retrieval

Ask a Language Model for Lottery Numbers: Concentration in Repeated Six-of-49 Outputs

We evaluate six language-model configurations on requests for six distinct random integers from 1-49. Across 1,200 attempted calls using four English prompt variants, 1,184 responses yielded valid tickets. Effective diversity of number frequencies ranged from 9.9 to 18.0, compared with simulated fifth-percentile thresholds of 46.6-46.7 under independent uniform six-of-49 sampling at the corresponding sample sizes. Systems produced 8-93 distinct unordered tickets, and their modal tickets accounted for 22.5-68.0% of valid responses. Two archived Polish Lotto samples provided a physical-lottery comparison, with effective diversities of 41.1 and 41.4 at smaller sample sizes. These results demonstrate substantial concentration under the tested deployment settings. They do not identify its mechanism or establish performance under other prompts, temperatures, or tool configurations.
Read the full abstract

Data Analysis and Statistics

Continuous Event Weighting for Resonance Searches

We develop a continuous event weighting framework for resonance searches that uses auxiliary event information without dividing the data into increasingly fine sensitivity categories. A hard selection is a binary weight, finite categories correspond to piecewise constant weights, and in the background dominated limit their fine partition approaches a continuous signal to background density ratio weight. The construction retains auxiliary discrimination in a single physically interpretable resonance spectrum, with calculable statistical variance and direct recovery of the selected signal yield. One million fully explicit event level pseudoexperiments using Gaussian, Gamma, and Beta auxiliary observables verify yield recovery, uncertainty propagation, and frequentist coverage while substantially improving discovery sensitivity. The method is further tested on publicly available CMS Run2016G and Run2016H diphoton Open Data as an independent real data benchmark, using detector level gluon fusion Higgs simulation only as the signal reference. This is a methodological test of the framework, not a CMS Higgs analysis. The results demonstrate a practical route to exploiting continuous auxiliary information without proliferating sensitivity driven event categories.
Read the full abstract

Quantum Physics

Quantum Query Advantage Requires Space

Hao, Huang, and Liu (STOC'26) recently showed that optimal quantum query complexity may require large workspace even for short-output problems, and asked whether a quantum query advantage over classical computation can itself require space. We resolve this question by exhibiting an explicit total Boolean function with quantum query complexity $Q=\widetilde{\Theta}(M)$, and randomized query complexity $R=\Theta(M^{21/20})$, whereas, for every fixed $0<\eta<1/100$ and $S\le O(M^{1/100-\eta})$, its $S$-space quantum query complexity satisfies $Q_S=\omega(M^{21/20})$. Consequently, $$ Q<R<Q_S, $$ so the unrestricted quantum query advantage disappears under sufficiently small workspace. The separation is obtained through a one-bit filtered-parity construction and a space-sensitive quantum lower bound based on compressed-oracle capacity and a new parity-to-capacity inequality, which may be of independent interest.
Read the full abstract

Combinatorics

Rooted spanning forests in wheel graphs: Fibonacci-Lucas formulas and extremal root configurations

In this paper, we study rooted spanning forests in the wheel graph $W_{N+1}$. For a root set $R$ consisting of rim vertices, we give an explicit formula, in terms of Fibonacci and Lucas numbers, for the number of rooted spanning forests in which each connected component contains exactly one vertex of $R$. When the central vertex is also included in the root set, we obtain a simple product formula involving only Fibonacci numbers. Furthermore, by using the correspondence between rooted spanning forests and vertex identification, we derive explicit formulas for the number of spanning trees of quotient graphs of wheel graphs obtained by identifying several vertices into one vertex. We then rewrite the obtained formulas combinatorially in terms of the numbers of matchings in path graphs and cycle graphs. In addition, for a fixed number $r$ of rim root vertices, we express the sum of the numbers of rooted spanning forests over all rim root sets of size $r$ as coefficients of generating functions, and we solve the extremal problem for the arrangement of roots.
Read the full abstract

Machine Learning

Backdoors Leave Structural Traces: FedMAST for Backdoor Detection and Containment in Federated Learning

Federated learning enables distributed training without requiring clients to share their raw data. However, its reliance on the integrity of the client-submitted updates exposes the global model to stealthy backdoor poisoning. Existing defenses often rely on individual evidence sources, but stealth-constrained attacks can adapt to these signals. Such attacks can suppress anomaly signals they are optimized to evade, yet their poisoned updates still leave residual structural traces. We propose FedMAST, a Federated Multi-Axis Structural Tracing defense for backdoor detection in federated learning. FedMAST scores client updates using complementary structural, spectral, and historical evidence and then applies tiered filtering and round-level containment to limit adversarial influence. To capture traces that isolated signals may miss, FedMAST uses squeeze-pair coherence scoring to expose coupled feature distortions and signed spectral-drift tracking to reveal persistent directional changes over time. Across six backdoor attacks, FedMAST achieves lower attack success rate (ASR) than baseline defenses in all nine evaluated comparisons, averaging 1.51% ASR and 94.84% main-task accuracy (MTA) across the complete 200-round runs. Over the full 200-round method-aware CovertLayers run, FedMAST achieves 1.53% ASR and 92.26% MTA, compared with ASRs of 100.00%, 99.67%, 99.53%, and 32.84% for FedAvg, MultiKrum, AlignIns, and FLAME, respectively.
Read the full abstract

Human-Computer Interaction

Beyond the Lab: Large-Scale Remote Cybersickness Research in Virtual Reality Using the VERA Platform

Cybersickness remains a major barrier for the adoption of virtual reality (VR), yet most existing knowledge is derived from laboratory-based studies with relatively small and homogeneous participant samples. In this paper, we investigate whether remote VR studies can produce cybersickness findings comparable to traditional in-lab experiments while enabling larger and more diverse participant populations. Using the Virtual Experience Research Accelerator (VERA), we deployed a remote adaptation of the standardized Cybersicker testbed for cybersickness research and collected data from N=263 participants using their own consumer VR headsets. We compared these results against a previously published in-lab dataset and a demographically matched subset of the remote sample. Across cohorts, cybersickness outcomes were consistent in direction and temporal pattern, including symptom onset trajectories, Fast Motion Sickness Scale (FMS) ratings, and Simulator Sickness Questionnaire (SSQ) responses, supporting the validity of remote cybersickness human-subjects research. Leveraging the larger remote dataset, we additionally examined demographic and individual-difference factors associated with cybersickness. These analyses confirm at scale the effects of sex, sickness susceptibility, and sickness expectation reported in smaller laboratory samples, and they add well-powered evidence on the contested relationship between age and cybersickness. These findings demonstrate that the VERA platform provides a viable means of remotely replicating laboratory-based studies and highlight its ability to support large, demographically diverse participant samples. This work establishes a validated methodological foundation for future large-scale remote studies of cybersickness and other VR research topics.
Read the full abstract

All authors Matt Gottsacker, Gerd Bruder, Daniel Zielasko, Alexander Giovannelli, Ali Haskins, Corey Clements, Chloe Beato, John Murray, Robert W. Lindeman, Rui Xie, Jonathan Beever, Nicholas Alvaro Coles, Valerie Jones Taylor, Tabitha Peck, Jeremy Bailenson, Gregory F. Welch

Instrumentation and Detectors

Kinematic Fitting of electromagnetic calorimeter data - an improved method

Kinematic fitting is widely used in particle physics experiments as a powerful tool to improve experimental resolutions, to suppress background and to provide selection criteria for the identification of specific reactions. Kinematic fitting methods, however, typically assume that the fitted quantities follow Gaussian distributions; an assumption which does not always hold. This is particularly true for energy measurements from electromagnetic calorimeters. This issue can largely be overcome by performing a transformation of the non-Gaussian variables into variables which follow a Gaussian distribution. The new adapted kinematic fitting procedure allows to better account for the non-Gaussian nature of the measured calorimeter energies, at the same time improving the accuracy and robustness of the kinematic fit. This leads to improved pull and confidence-level distributions of the kinematic fit as well as improved invariant mass distributions and signal-to-background ratios in the final event sample.
Read the full abstract

Computation and Language

H2LooP Telecom Model v1: From Telecom Comprehension to Autonomous Issue and PR Resolution

We present H2LooP Telecom Model v1, a domain-specialized large language models fine-tuned for the telecommunications industry. We release two domain-adapted model variants serving complementary use cases: a comprehension-focused variant for telecom domain question answering and reasoning, and an agentic variant for autonomous telecom code generation, pull request resolution, and code commits on production repositories. H2LooP Telecom achieves strong results on the GSMA Open Telecom Lite (OT-Lite) benchmark and a proprietary telecom code generation benchmark, outperforming frontier closed-source models such as GPT-5 and Claude Opus on independent leaderboard evaluation, while preserving general-purpose capabilities. The Comprehension variant achieves 81.8% weighted average on OT-Lite Pass@3, and, independently, ranks 5th overall on the official community-run Open Telco AI Leaderboard* at only 31B parameters-ahead of frontier closed-source systems including Claude Opus 4.6, GPT-5, Gemini 3 Flash, Grok-4-fast, and Kimi K2.5. Our agentic variant obtains a relative improvement of +8.8% in AST Similarity and +20.0% in Location IoU over the base model on telecom code generation, while maintaining identical MMLU (74.0%) and BFCL v3 multi-turn function calling (79.0%) performance, indicating zero catastrophic forgetting. Domain specialization on curated telecom corpora, spanning 3GPP standards, O-RAN specifications, network telemetry, and real repository commits, yields substantial improvements over general-purpose models of equivalent scale, approaches frontier closed-source models on domain-specific evaluation, and is independently corroborated by our official leaderboard standing.
Read the full abstract

All authors Amit Singh, Vedant Nipane, Mayank Goel, Pulkit Agrawal, Sairanjan Mishra

Machine Learning

Statistical Inference for Adversarial Training: Central Limit Theorems via Optimal Transport

The purpose of this paper is to rigorously quantify the statistical and learning-theoretic properties of adversarial training models for classification. Equivalently, we establish the statistical properties of empirical optimal partial transport. Precisely, first we provide two types of central limit theorems (CLT): CLT centered at the expected empirical value, and CLT centered at the population one with smoothing. These results are based on the uniqueness of optimal potential for various equivalent optimal transport formulations, and the empirical process theory argument. For the binary setting, we indeed prove the uniqueness of optimal potential by leveraging the connection between optimal partial transport and the derived multi-marginal optimal transport formula. As byproducts, we also obtain the stability of a saddle point of the adversarial training model, and the sample complexity and concentration probability of the generalization error.
Read the full abstract

Computation and Language

Knowledge Graph-Augmented Ambient AI for Clinical Note Generation

Ambient AI is increasingly adopted in healthcare to automatically generate clinical notes from patient-clinician conversations, with the potential to substantially reduce clinician documentation burden. However, generated notes may omit clinically relevant information discussed during the encounter, creating information gaps that can affect downstream care. Knowledge graphs (KGs) constructed from encounter transcripts can provide a structured representation of what was discussed and enable systematic identification of missing information from generated notes that are critical for patient care. In this study, we introduce Coverage-Directed Revision (CDR), a model-agnostic framework that constructs a KG from the encounter transcript, identifies medical concepts absent from an initially generated note, and directs large language models (LLMs) to restore the missing information without modifying the underlying note-generation system. We evaluate CDR on two datasets: 1) Pitt-Bench, a local dataset comprising rehabilitation sessions, and 2) ACI-Bench, a public dataset for benchmarking clinical note generation. We tested four underlying LLMs widely used in ambient AI systems. The results show that CDR consistently improves content recall across all evaluated conditions. Our study provides a practical approach for improving the completeness of ambient AI-generated clinical documentation.
Read the full abstract

All authors Jakir Hossain, Yi-Fei Zhao, Hongjian Wang, Minmei Shih, Katie Leigh Mullen, Ahmad P. Tafti, Leming Zhou, Manoj Purohit, William Hogan, Jay Zeng, Elizabeth Skidmore, Yanshan Wang

Machine Learning

Task-Aware QUBO Allocation for Mixed-Precision Quantization

Mixed-precision quantization requires discrete allocation of weight and activation bit-widths, followed by recovery of the selected network. We develop a task-aware quadratic unconstrained binary optimization (QUBO) surrogate with separate weight and activation profiles, a bit-operation (BOP) cost, and selected structural priors. QUBO provides a network-wide allocation that can be refined through direct validation-based PROTES search. On a compact NAFBlock-based denoiser, the refined route achieves 37.192 dB after LSQ+ at 4.035% routed-layer BOPs, versus 37.092 dB at 4.101% for a HAWQ-style baseline. The repeated-search primary experiment shows that LSQ+ largely closes the quality gap between QUBO allocation and expensive direct refinement. An additional restoration architecture retains a larger recovered gain, indicating that refinement's value depends on architecture and recovery. We evaluate quality, achieved cost, routing stability and optimization expense together. Deployment measurements characterize a fake-quantized floating-point implementation; BOP reductions describe analytical allocation savings.
Read the full abstract

All authors Osama Orabi, Artur Zagitov, Hadi Salloum, Viktor A. Lobachev, Yaroslav Kholodov

Medical Physics

An interpretable model of spectral scattering of arterial pulse waves in the circle of Willis encodes occlusion location

Background and objective. Carotid Doppler ultrasound is the most widely available bedside probe of cerebral haemodynamics, and machine-learning classifiers fed carotid velocity spectra can localise intracranial occlusions—but at the price of black-box models, thousands of training samples, and fragility to anatomical variants. We ask whether the same information can be obtained from an interpretable physical model. Methods. We formulate a linear frequency-domain one-dimensional model of a 26-segment circle-of-Willis network in which each cardiac harmonic propagates on a Womersley transmission line and is scattered independently by a lesion. The model is validated against a nonlinear one-dimensional solver and three-dimensional stenosis computations at two severities, coupled to a physics-based synthetic spectral-Doppler pipeline, and used to generate a 9{,}600-case virtual cohort for occlusion-localisation classification with physics-guided features. Results. Mean flow divisions agree with the nonlinear solver to three significant figures, harmonic magnitudes within a ratio of 0.90–1.04, and carotid waveforms to 6.1\,% in relative $L_2$ norm after a single kinematic area correction. The benchmark reproduces healthy flow splits within 1–9\,%, collateral-channel flows within 11–16\,%, and the collateral-pathway hierarchy. Physics-guided harmonic-ratio features localise twelve lesion classes at 95.1\,% accuracy from five training samples per class—over an order of magnitude fewer than waveform-driven convoluted neural networks; augmentation partially restores robustness to measurement noise and anatomical variants. Conclusions. The framework explains the physical origin of the empirically discriminative 2–12\,Hz Doppler band and makes occlusion localisation interpretable and sample-efficient.
Read the full abstract

Computer Vision

Evaluation of MLLM-Agnostic Plug-and-Play Keyframe Selection Methods for Long Video Understanding

Multimodal large language models (MLLMs) cannot process every frame of a long video because of limitations in visual-token and computational budgets. Three primary approaches have been proposed to enhance their long-video understanding capabilities: (i) Retraining an MLLM on a large video corpus and/or extending its input length; (ii) Training an adapter for a specific MLLM that takes the entire video and the query as input and selects the most relevant video frames; and (iii) Developing a training-free, plug-and-play (PaP) adapter that is MLLM-agnostic. We refer to the third approach as PaP keyframe selection. A PaP method may use only candidate video frames without considering the query, or it may use both candidate video frames and the query. The first approach is prohibitively expensive. The second approach requires substantial training time and computational resources, but it is accessible to many because an adapter contains significantly fewer trainable parameters than an entire MLLM. The third approach has the lowest computational cost and is therefore broadly accessible. To the best of our knowledge, only five PaP methods have been reported within the past year. All of these methods have been evaluated on one or more video question-answering benchmarks and have demonstrated improvements in long-video understanding. However, the methods were evaluated on different benchmarks using different MLLMs. We present a comprehensive evaluation of these five methods using three MLLMs across three long-video understanding benchmarks. Our results show that QAaF achieves the best performance in 13 of the 15 aggregate evaluation settings, while FOCUS ranks second overall. These results provide a common experimental reference for comparing training-free keyframe-selection methods for MLLMs.
Read the full abstract

Tissues and Organs

Interstitial flow in the chick yolk sac exhibits organ-scale patterns driven by segregated leakage and drainage

Interstitial flow plays a key role in drug delivery, angiogenesis, cancer, and edema. Recent studies have identified connected interstitial pathways that can transport material over long distances. However, the spatial scale of physiological interstitial flow remains unclear: despite the existence of connected pathways, flow may be dominated by nearby vascular filtration or extend over longer distances. Here, we identify organ-scale interstitial flow patterns in the chick yolk sac and elucidate the mechanisms that determine their spatial scale. The yolk sac contains an organ-scale interstitial region that enables large-scale flow patterns to be identified without truncation. Our approach combines experimental imaging with computational modeling of coupled blood and interstitial flow in the whole yolk sac. Scaling analysis identifies a hydraulic conductivity ratio that controls the transition from small-scale to organ-scale flow. In organ-scale patterns, we find interstitial flow speed to be substantially greater than the transvascular flow speed. Mechanistically, the organ-scale patterns arise from the spatial segregation of the leakage and drainage of interstitial fluid from the vessels. These findings have important implications for biological transport by interstitial flow, suggesting that flow-mediated cues may be sensed locally but generated nonlocally.
Read the full abstract

All authors Adithya Srinivasan, Deshik Reddy Putluru, Elizabeth A. V. Jones, Hector Gomez

Signal Processing

Robust Wave Origin Detection from Sensor Array Data via Randomized Hough Transform and Model Fitting

The robustness of methods for characterizing the origin of wave-like propagating signals is often challenged by the heterogeneity of transmitting media and sensor network. Detecting an individual wave under such noisy conditions remains particularly difficult. In this paper, we propose a robust framework to estimate wave origins from spatiotemporal data collected by a sensor array. First, a Butterworth filter is designed to extract relevant signals from each sensor channel. Next, a 3D randomized Hough transform is introduced to identify candidate signals originating from the same wave front. Finally, a wave propagation model is fitted to the identified signals using the least-squares method. By combining the iterative voting mechanism of the randomized Hough transform with least-squares optimization, our proposed method demonstrates superior robustness and computational efficiency in both simulation studies and real-world data analyses.
Read the full abstract

Multiagent Systems

Machine learning-assisted calibration of Agent-based Models: surrogate-based optimization with Genetic Algorithm and Particle Swarm Optimization

Calibrating an agent-based model (ABM) is difficult because its objective landscape is stochastic and rugged, and can be evaluated only through costly black-box simulations. This study adapts inner-loop surrogate-assisted evolutionary computation (SAEC) to ABM calibration by embedding a machine-learning surrogate within genetic algorithm (GA) and particle swarm optimisation (PSO). At each iteration, the surrogate screens the candidates and the simulator validates only the top 50%, reducing simulation demand while correcting surrogate errors. We evaluate a full factorial of 48 configurations combining two optimisers, five surrogates, and four calibration objectives on two contrasting ABMs, the Brock-Hommes asset-pricing model and the Island growth model. Parameter recovery is measured against pseudo-true values. Relative to the strongest pure-optimiser baseline, the best ML-assisted configurations reduce RMSE by 20.0% on Brock-Hommes and 63.8% on Island, while reducing computation time by 32.1% and 61.1%, respectively. ANOVA with Dunnett's post-hoc tests confirms significant time savings for every surrogate under both Brock-Hommes optimisers and under GA on Island. No surrogate differs significantly from the pure-optimiser baseline in parameter-recovery accuracy on Island; the reported accuracy gains are therefore best-configuration outcomes rather than average effects resolved at this sample size. The best surrogate-optimiser-objective combination changes with ABM complexity, indicating that surrogate-assisted calibration depends on the model and does not admit a universal default recipe.
Read the full abstract

Computer Vision

Pixel-wise Planarity for High-Precision Monocular Plane Segmentation

Plane segmentation from a single RGB image remains challenging due to imprecise region grouping and geometrically inconsistent supervision, often leading to over-segmentation and false planar detections. We propose instead a pixel-wise planarity prediction framework for robust monocular plane segmentation. Building on a pretrained monocular geometric backbone predicting depth and surface normals, we introduce a dedicated planarity head that estimates per-pixel planarity confidence. During inference, predicted depth, normals, and planarity are combined in a lightweight region-growing procedure that enforces geometric consistency when forming plane segments. We further analyze existing plane ground-truth annotations and demonstrate substantial geometric inconsistencies under strict distance thresholds. Across multiple datasets, our method achieves improved geometric precision and segmentation quality compared to prior state-of-the-art approaches, while improving computational efficiency. Our code and models are available at https://github.com/alpayozkan/PixelwisePlanarity.
Read the full abstract

All authors Ahmetcan Yavuz, Alpay Ozkan, Rémi Pautrat, Shaohui Liu, Marc Pollefeys

Computer Vision

SJD-SV: Speculative Jacobi Decoding with Semantics Verification for Autoregressive Image Generation

Speculative Jacobi Decoding (SJD) is an important approach for accelerating autoregressive image generation. Although SJD has shown superior performance, recent studies point out that it usually suffers from a token ambiguity issue during token verification but its reason can not be well explained. To figure out this reason, in this paper, we conduct a visualization analysis on vision token and find that different from text tokens, vision tokens generally corresponds to some local, small, and unclear vision details, which means only using single token is difficult to accurately express a certain semantic, thereby causing token ambiguity issue. To this end, we propose a novel Speculative Jacobi Decoding with Semantics Verification (called SJD-SV), for accelerating autoregressive image generation. The key idea is that leveraging the strong correction characters between tokens to recognize semantic-aware token subsequence and then instead of perform token-by-token verification, turning to perform verification on semantic-aware token subsequence level for accelerating image generation. In particular, our method is plug-in, which can be directly integrated into existing SJD and its variants. Extensive experiments on various datasets show that existing SJD methods achieve significant performance improvement after integrating our SJD-SV method.
Read the full abstract

All authors Baoquan Zhang, Bingqi Shan, Shihao Fang, Kenghong Lin, Xutao Li, Yunming Ye