Machine Learning
Monitoring Urban Traffic Dynamics at Fine Spatiotemporal Resolution Using Distributed Acoustic Sensing and Deep Learning
Hao Tian, Heng Cai, Xiaowei Chen et al. (+1 more)
Mapping the distribution of traffic dynamics at high spatiotemporal resolution is a fundamental question in transportation research. Distributed acoustic sensing (DAS), an innovative seismic observation tool, emerges as a promising solution for real-time urban traffic monitoring at high spatial and temporal scales. Distributed acoustic sensing repurposes existing underground fiber-optic cables as dense, continuous sensor arrays, enabling passive and
privacy-preserving monitoring of roadway traffic activity at meter-level spatial and second-level temporal resolution. This study examines whether integrating DAS and deep learning models can serve as a continuous and efficient urban traffic observatory for revealing urban traffic dynamics (i.e. traffic volume and congestion, event-driven changes) at high spatiotemporal resolution. Using a DAS deployment along a roadway network in the City of College Station, Texas, USA, this study develops a deep learning-empowered analytical framework that converts raw ground vibration waveforms into spatiotemporal representations, detects vehicle trajectory, and infers traffic states from aggregated traffic volume and speed. A hybrid training strategy combining synthetic and manually annotated DAS images is used to improve vehicle detection under noisy and congested conditions, with model outputs further aggregated to characterize system-level traffic dynamics.
Read the full abstract ▾
All authors Hao Tian, Heng Cai, Xiaowei Chen, Yifan Yang
Machine Learning
Vector Bellman Theory for Multichain Robust Average-Reward Markov Decision Processes
Yue Wang and George Atia
Robust average-reward Markov decision processes provide a fundamental framework for long-term performance optimization under uncertainty, and can have optimal long-run rewards that depend on the initial state. This state dependence requires a vector Bellman theory that accounts for both recurrent-class rewards and transition uncertainty. We develop such a theory for finite models with compact, post-action $(s,a)$-rectangular ambiguity. A gain-first, bias-second optimization principle yields a coupled vector gain-bias system, and every finite solution identifies the optimal robust gain and supplies stationary saddle strategies against history-dependent opponents, simultaneously from all initial states. We further characterize solvability through stationary gain conditions and a uniform bound on canonical transient corrections, and give sufficient conditions that permit distinct recurrent-class gains. The certificates also yield asymptotically affine trajectories of the robust Bellman operator, based on which we design a robust approximately shifted Halpern planning
algorithm. Under finite Bellman solvability, the gain estimates and Bellman displacements converge to the optimal gain vector, and every extracted greedy controller is average-optimal after a finite, instance-dependent budget. These results thus connect finite Bellman certificates to undiscounted planning for state-dependent robust average rewards, providing theoretical understandings.
Read the full abstract ▾
Discrete Mathematics
Packing Tails of Reciprocal Rectangles into Squares of Equal Area
Yu Jiang
The Meir–Moser rectangle-packing problem asks whether all rectangles with side lengths \(1/n\) and \(1/(n+1)\), for \(n\ge1\), can be packed into the unit square with pairwise disjoint interiors. We establish a tail version of this problem. Let \(R_n\) denote the rectangle with these side lengths. We prove that there exists an integer \(m_0\) such that, for every \(m\ge m_0\), the family \(\{R_n:n\ge m\}\) admits a packing, by translations and right-angle rotations, into a square of side length \(m^{-1/2}\), with pairwise disjoint interiors. The area of the square equals the sum of the areas of all the rectangles. The
geometric construction recursively decomposes rectangular gaps, while local randomized quotas and random permutations assign subsequent integer indices. We separately control the total area of waiting gaps and the assignment load at each index. The proof is organized in six steps: a finite-prefix reduction,
geometric row decompositions, an area bootstrap, a sharp source-load estimate, control of the actual adaptive construction, and a compactness limit. For every finite time horizon, the probability of failure has a bound that is independent of the horizon and can be made arbitrarily small. The adaptive step uses a permanent load ledger, actual fresh height queries, and a one-sided comparison with a frozen source experiment. Compactness then yields an infinite packing. The final packing statements have been checked in Lean 4. A sufficient threshold is \(m_0=10^{1000}\). This result applies only to sufficiently late tails and does not resolve the original Meir–Moser rectangle-packing problem for the full sequence starting at \(n=1\), which remains open.
Read the full abstract ▾
Sound
Statistical Models for Automatic Fingering-Annotated Piano Sheet Music Transcription
Daniel Penner and Abram Hindle
Machine learning tools have significantly aided automatic piano music transcription; however, this domain has focused primarily on accurately predicting the pitches and timings of played notes. To produce sheet music for the piano, notes must be separated into two staves, one for each hand, and good sheet music often contains fingering annotations to guide the player when sight-reading or learning fast or complex pieces. We propose 8 statistical approaches for combined hand and fingering annotation of transcribed piano notes, including baseline hidden Markov models, rule-based methods, a synthesis of existing approaches, and N-gram
language models. Furthermore, we develop a pipeline for complete transcription from piano audio to fingering-annotated sheet music. Evaluations with the PIG
dataset demonstrate that our Synthesis model achieves a hand separation accuracy of 90.8% and a joint hand and finger annotation accuracy of 56.6%. These approaches serve as a new baseline for further research into this problem, while our pipeline demonstrates the feasibility of a combined system for automated note transcription, hand separation, and fingering annotation.
Read the full abstract ▾
Systems and Control
Optimal Measurement Selection for Certifiable Voltage Monitoring in Power Distribution Systems
Peng Zhang, Line Roald and Manish K. Singh
Maintaining consumer-end voltages within mandated limits is a central task in operating power distribution systems. Rising penetration of distributed generation, electric vehicles, and flexible loads complicates this task, while a scarcity of real-time measurements leaves distribution networks largely unobservable. Operators commonly monitor a few bellwether meters, which indicate operational health but certify nothing about the unmeasured buses. This work asks how to select measurements that are reported in real time and can certify network-wide voltage-limit compliance. We first illustrate the geometry of the measurement values that certify voltage safety at unmeasured nodes. Building on this geometry, we formulate a bilevel optimization that selects measurements to maximize certification capability, introducing a novel safety-violation metric with favorable monotonicity properties. We derive a strong-duality-based single-level reformulation and two scenario-reduction schemes with quantifiable suboptimality. Numerical tests on the SCE 56-bus system corroborate the superior certification capability and computational efficiency of the proposed approach.
Read the full abstract ▾
Computation and Language
Script Choice in LLMs: Evidence for Late-Layer Commitment
David Kletz, Sandra Mitrović, Itay Sabato et al. (+2 more)
In this paper, we investigate how script knowledge is distributed across the layers of LLMs using two complementary interpretability methods: logistic regression probing and logit-lens analysis. Our probing experiments reveal a clear asymmetry: both the input script and the instructed output script are encoded in the earliest layers of the network, while, in contrast, commitment to the actual output script emerges only in the final layers, with the model's intermediate representations defaulting to Latin throughout most of the layers. This two-stage process is confirmed by logit-lens analyses, which show that script commitment consistently occurs at the very last layers of the LLMs. Together with the weaker script-following performance observed in smaller models, these results form a converging body of evidence linking script commitment to model depth, with broader implications for the design of sufficiently deep, inclusive
multilingual architectures.
Read the full abstract ▾
All authors David Kletz, Sandra Mitrović, Itay Sabato, Ljiljana Dolamić, Fabio Rinaldi
Digital Libraries
Residential Electricity Consumption Dataset for Sri Lanka (RECON-SL)
Chanuka Algama, Merl Chandana, Isuruni Fernando et al. (+6 more)
This article describes the Residential Electricity Consumption
Dataset for Sri Lanka (RECON-SL), a multi-source resource that integrates utility records, high-frequency smart meter data, and household surveys from the service area of Lanka Electricity Company (LECO) in Sri Lanka. The
dataset captures electricity use for 4,063 households, collected over the period October 2022 to January 2025, with coverage at monthly, 6-hour, and 15-minute intervals. Monthly billing-cycle data provide complete coverage across all households, while smart meter data are available for a subsample of 1,438 households, offering both fine-grained 15-minute readings and coarser 6-hour records. Three survey waves complement these consumption measures, collecting information on demographics, housing, appliance ownership, and perceptions of energy security. Across its components, the
dataset includes more than 50 million meter readings and detailed survey responses from over 11,000 household interviews. As the first
dataset of its kind in Sri Lanka, and among the few worldwide to link smart meter records with longitudinal survey data at this scale, it provides a unique resource for energy, machine learning and policy research, with known gaps in smart meter coverage documented for users. The
dataset can support diverse applications, including electricity load
forecasting, socio-economic and behavioral energy research, distribution planning, and policy analysis in energy affordability and equity. Access is provided through an open repository, with accompanying documentation and scripts to facilitate reuse.
Read the full abstract ▾
All authors Chanuka Algama, Merl Chandana, Isuruni Fernando, Dinithi Dissanayake, Amanda Ariyaratne, Nipuni Habaragamuwa, Pasindu Ranage, Jesús Ramos, Kasun Amarasinghe
Machine Learning
The Mechanics of Delta Learning: Target Design for Generalizable Scientific Machine Learning
Kareem M. Gameel, Ihor Neporozhnii, Sjoerd Hoogland et al. (+1 more)
In scientific machine learning, $\Delta$-learning trains models on residual errors relative to physical baselines, assuming that more accurate baselines with smaller residual scales inherently improve downstream performance. Here, we demonstrate that residual scale alone is an insufficient heuristic for learnability. Evaluating molecular graph neural networks on total energy targets, we show that complex local descriptor baselines can yield small residual targets that are disproportionately rough within architecture-informed proxy spaces and harder to learn relative to their scale. Conversely, semi-empirical baseline reduces both scale and normalized roughness, improving in-domain and out-of-domain prediction. We introduce scale-normalized graph Dirichlet roughness ($D_{\text{IQR}}$) as a pre-training diagnostic for residual learnability and establish baseline complementarity as a core target-design principle, elevating target space formulation alongside model architecture as a key axis for scientific machine learning.
Read the full abstract ▾
All authors Kareem M. Gameel, Ihor Neporozhnii, Sjoerd Hoogland, Oleksandr Voznyy
Data Structures and Algorithms · Cross-listed from math.PR
Eigenvalue and Eigenvector Approximation for Random Matrices Using Low-Degree Polynomials
Yihan Zhang
We initiate the study of approximating the top
eigenvalue and eigenvector of a random symmetric
matrix $ A \in \mathbb{R}^{n\times n} $ using $ q(A)b $ where $q$ is a degree-$d$
polynomial and $b$ is a standard Gaussian vector independent of $A$. For spiked GOE $ Y = \lambda vv^\top + X $, we identify $ d_\star = \frac{\log(n)}{2\log(\lambda)} $ to be the critical degree threshold above which accurate approximation of the top
eigenvalue and eigenvector is possible. This sharpens the common belief that spectral methods can be implemented by $ O(\log(n)) $-step power iterations and offers a precise connection between spectral methods and low-degree
polynomial algorithms, a popular proxy for all
polynomial-time
algorithms. For GOE $X$, we identify $ d_\star = n^{1/3+o(1)} $ to be the critical degree threshold for top eigenvector approximation, whereas constant degree suffices for top
eigenvalue approximation. Moreover, in the limit where $ d/n^{1/3} $ converges to a positive finite constant, we compute the exact asymptotic eigenvector approximation accuracy in terms of the expected squared overlap. These results significantly improve upon predictions made in randomized numerical linear
algebra for deterministic data
matrices that the iteration count of power methods with random initialization is governed by the inverse spectral gap. Technically, our analyses leverage extremal properties of Chebyshev
polynomials and draw upon the rich literature of random
matrix theory.
Read the full abstract ▾
Distributed Computing
Where Does Streaming State Cost Go? A Reproducible Comparison of Flink and Kafka Streams on Kafka
Kiran N Kumar and Santhosh Kumar Saminathan
This paper presents a controlled comparison of exactly-once Kafka pipelines implemented with Apache Flink and Kafka Streams, two engines with different state-management architectures. The state management in Flink occurs through checkpoints in external storage, whereas Kafka Streams restores local state by replaying broker changelogs. The experiments evaluate the effects of the two designs on latency, resource consumption, configuration sensitivity, and recover from injected failures. The engines produced expected outputs across 50 correctness trials, five for each engine-workload combination (exact 95% CI 0.929-1.000). The latencies, however, showed differences when tested in 30 minute trials at a fixed rate of 100 events/sec. The measured ingestion to output interval was 4-6 ms for stateless workloads and 4.3-6.2 s for windowed workloads. The median p99 was 1.97-1.97s for stateless workloads and 5.68-9.92s for windowed respectively. Further, a sensitivity study showed that increasing the durability interval from 1,000 to 10,000 ms shifted the latency distributions in Kafka Streams and Flink between stages. Kafka Streams' W3 median p99 T2-T1 increased from 6,085 to 23,155 ms; Flink's T3-T2 increased from 732 to 8,702 ms. Failure injection experiments demonstrate that latency can interfere with incomplete processing behavior. The results show that exactly-once correctness is a necessary but insufficient measure of stream processing performance. Kafka Streams completed 0/5 stateful JVM-kill and 1/5 in each stateful local-volume-loss trials, while corresponding Flink
cells completed 5/5. The cost, placement and magnitude depend on state management, architecture, configuration, and measurement location. Accordingly, evaluations of fault-tolerant streaming systems should holistically measure latency, completion, recovery behavior, and correctness.
Read the full abstract ▾
Sound
Reward-Tilted On-Policy Distillation for Acoustic Grounding in Audio-Language Models
Kaiyang Li, Shaobo Han, Yue Tian et al. (+1 more)
Audio-
language models (ALMs) can exploit textual shortcuts to answer questions while overlooking acoustic evidence, weakening audio understanding. On-policy
distillation (OPD) trains compact ALMs by supervising student-generated responses with teacher predictions, but does not explicitly distinguish acoustic support from linguistic predictability. We propose Reward-Tilted On-Policy
Distillation (RT-OPD) to strengthen acoustic grounding. Given the same question and student-generated text, a frozen teacher predicts the next
token with and without audio inputs. Their log-probability contrast defines a reward that reshapes the teacher distribution for reverse-KL
distillation, emphasizing the additional evidence provided by audio. Across two compact students and three
benchmarks, RT-OPD consistently outperforms Vanilla OPD. Experiments with silenced and replacement audio further suggest that RT-OPD strengthens the student's reliance on acoustic evidence. Our 3B model achieves 72.72% accuracy on MMAU, the highest among the compared 3B models and competitive with several 7B and 8B models. Code and model checkpoints are available at https://github.com/KaiyangLi1992/RT-OPD.
Read the full abstract ▾
All authors Kaiyang Li, Shaobo Han, Yue Tian, Shihao Ji
Human-Computer Interaction
Data Patching
ATM Mizanur Rahman, Syed Ishtiaque Ahmed and Sharifa Sultana
To provide data-driven quality services to their citizens, every institution determines the acceptability of the citizen data and datafied mechanisms through institutional protocols, including preset standards and policies. However, data meeting standards at one institution might fail to meet a different institution's standards in subsequent phases of the intended citizen services due to mismatched protocols, leading to data devaluation in the citizen service ecosystem. Terming this phenomenon cross-institutional data devaluation, we investigate its causes and workarounds through interviews with 41 Bangladeshi participants. We found that traditional data
auditing mechanisms cannot solely address data devaluation; hence, we draw on our findings and theory of cross-institutional AI audits to propose the Citizen-centered Cross-institutional Data Audit (CCDA). We also discuss design and policy implications of CCDA in HCI and datafied citizen services.
Read the full abstract ▾
Artificial Intelligence
Learned Cross-Task Relationships in Multi-Task Models
Victor Zhang, Yiping Yuan, Florian Raudies et al. (+4 more)
We propose a framework that learns cross-task relationships in multi-task models by approximating the joint distribution of task labels through targeted pairwise relationships. This approach improves performance via transfer learning and enhances information extraction without the intractable complexity of modeling the full joint space. Although our framework applies to any multi-task system, we demonstrate its efficacy within YouTube's production recommendation systems. Experiments across the Notifications, Homepage, and Watch Next surfaces show improvements in both accuracy and user satisfaction metrics. Finally, we propose a workflow template to facilitate broader future implementation.
Read the full abstract ▾
All authors Victor Zhang, Yiping Yuan, Florian Raudies, Bosun Adeoti, Brian Y. C. Leung, Sanjay Surendranath Girija, Naijing Zhang
Machine Learning · Cross-listed from eess.IV
Physics-Guided Multi-Objective Deep Learning for Ultrasound RF Data Interpolation in Resource-Constrained Imaging
Luoyuan Zhang, Yiyang You, Ananya Tandri et al. (+4 more)
Ultrasound imaging increasingly targets portable, point-of-care, and wearable settings where constraints on power, bandwidth, and hardware complexity often necessitate sparse data acquisition in spatiotemporal scanning. However, image reconstruction using the sparse data can introduce insufficient phase information in coherent beamforming process, resulting in grating-lobe artifacts that degrade imaging contrast resolution. We present a physics-guided, data-driven framework for sparse-to-dense
radio-frequency (RF) reconstruction that aligns training with downstream image formation. Our approach trains an end-to-end interpolation network using a hybrid supervision scheme that combines an RF-domain and a beamforming-domain loss with exponential moving average (EMA) to stabilize the multi-objective training. To improve generalization under variable acquisition layouts, we also introduce a random-skip masking strategy that varies sparsity patterns during training so a single model can handle diverse decimation factors and irregular channel configurations. We evaluate the framework on a held-out test set using the mean structural similarity index measure (SSIM) between reconstructed and ground-truth beamformed images. Across decimation factors $\times 2$ to $\times 13$, the best-performing configuration maintains mean SSIM around 0.95. Overall, the results show consistent gains in RF reconstruction and post-beamforming image quality across diverse acquisition conditions. This approach enables robust, high-quality ultrasound imaging at resource-constrained settings by allowing more sparse scanning in spatiotemporal domain.
Read the full abstract ▾
All authors Luoyuan Zhang, Yiyang You, Ananya Tandri, Yinan Feng, Hyunwoo Song, Jeeun Kang, Youzuo Lin
Artificial Intelligence
Agent Memory with Episodic Retrieval for Financial Decision-Making
Nuoyue Xu, Jiang Liu, Wenxuan Huang et al. (+3 more)
Large
language models (LLMs) have demonstrated strong capabilities in financial analysis and
reasoning, inspiring recent advances in
agent-based trading frameworks. While these systems show promise, prior approaches either emphasize long-horizon
forecasting or operate as stateless analyzers, limiting their applicability to the demands of trading in complicated settings. To address these gaps, we introduce META (
Memory Enhanced Trading
Agent), the first RAG-like episodic-
memory-augmented multi-
agent framework for financial decision making. META integrates a family of specialized indicator
agents (e.g., Trend, MACD,
Stochastic, RSI, SMA, AVWAP, Heikin-Ashi) with a Decision
Agent that fuses their reports, and a
Memory module that retrieves and updates past trading episodes encoded as
market state
embeddings with outcomes and reflections. By recalling relevant experiences and adaptively reweighting signals under similar
market regimes, META achieves improved directional accuracy and robustness under short-horizon evaluation. Our results demonstrate that episodic
memory provides a powerful mechanism for regime-aware, interpretable, and low-latency decision-making in trading and decision making. The code of this project is released on GitHub.
Read the full abstract ▾
All authors Nuoyue Xu, Jiang Liu, Wenxuan Huang, Xiang Zhang, Juntai Cao, Jiaqi Wei
Hardware Architecture
Xtrace: High-Fidelity GPU Intra-Kernel Tracing via Binary-Level Instruction Splicing
Zhuobin Huang, Kai Zhang, Weihao Cui et al. (+5 more)
Modern GPU kernels fuse increasingly more work into a single kernel, and intra-kernel tracing has become the mainstream method to profile them. Tracing inserts probes into the kernel to record its runtime states, and the fidelity of the trace determines the efficiency of performance optimization. Unfortunately, existing tools insert probes before compilation. These tools interfere with the compiler's optimizations, so they trace a different binary from the one the GPU executes. They also add significant runtime overhead. Xtrace is the first GPU kernel tracing system with near-zero compile-time interference and minimized runtime overhead. Xtrace inserts probes directly into the compiled kernel binary. It reuses only the registers that hold dead values at the insertion address and resolves all hazards with the compiler's hazard tables. It further schedules the instruction order, register allocation, and control bits to minimize the runtime overhead the probe introduces. Xtrace supports 19 NVIDIA and AMD GPU architectures, and is publicly available for use at https://g-watch.github.io. We evaluate Xtrace on major production large
language model (LLM) kernels against the state-of-the-art tracers
Neutrino and IKET from NVIDIA. On H100, B300, and MI300X GPUs, Xtrace preserves 94-98% of the instructions of the kernel, while existing tools preserve only 8-48%. Xtrace adds only 0.9-2.8% overhead, while existing tools add 3.8-75.6%. Xtrace guides a coding
agent to reach the same FlashAttention-3 performance with 3.9x fewer iterations than existing traces do. Thanks to our binary-level instrumentation, Xtrace also traces the faster closed-source cuDNN kernel, which guides the
agent to lift the open-source FlashAttention-4 by 5.2-13.3% in throughput.
Read the full abstract ▾
All authors Zhuobin Huang, Kai Zhang, Weihao Cui, Hongshi Tan, Liang Luo, Christopher Dewan, Shen Li, Bingsheng He
Robotics
Human-in-the-Loop Geospatial Annotation for Rapid Dataset Construction in Field-Deployed UAV Systems
Morgan Masters, Adam Korycki, Nikolaas Bender et al. (+4 more)
Real-world perception systems must adapt to changing environments, but manual image annotation cannot scale to field data volumes. We present BirdsEye, which shifts expert annotation from images to the field: an operator records target locations in world coordinates using RTK positioning and calibrated projective geometry propagates each observation to all frames where the target is visible. To quantify how well physical annotations align with image observations, we derive a first-order mapping from camera-pose uncertainty to pixel uncertainty and validate it against Monte Carlo simulation. This mapping is linear in the six per-axis pose variances, so it inverts into a sensor design tool: we give a sufficient condition converting an annotation tolerance into a convex set of admissible pose-noise budgets, a closed-form largest admissible scaling of a deployed sensor suite, and a unique per-axis pose specification under an equal-budget-share allocation. We also analyze the planar-surface approximation underlying the projection, which holds up to 10 degrees of terrain slope. By direct measurement, we show that system projection accuracy is sub-decimeter (sub-30 pixel) at AGL altitudes of 10-20m under conditions excluding sustained yawing. During an in-field case study across three agricultural sites, two field workers produced 12,524 annotated frames carrying 55,600 labels in roughly 12 hours (25.5x per-worker rate increase over manual labeling). Detectors trained on imagery collected by this workflow recovered 56-89% of in-view surveyed targets at a geographically distinct farm, at pre-registered operating points; human review of the leading configuration estimates detection precision at 83-87%, spanning three tie-break conventions for clusters carrying contradictory human verdicts.
Read the full abstract ▾
All authors Morgan Masters, Adam Korycki, Nikolaas Bender, T. Luca Altaffer, Nick Kuipers, Colleen Josephson, Steve McGuire
Robotics
TAPESIM: Efficient Simulation of Adhesive Tape Dispensing for Robotic Manipulation
Zhaofeng Luo, Xinyu Lu, Jaehoon Choi et al. (+10 more)
Applying adhesive tape to secure wire harnesses or seal packages requires
robots to coordinate a flexible strip, a moving roll, and surfaces that attach and detach. Simulation could make these interactions repeatable for
robot development and evaluation, but resolving every adhesive layer is expensive and can suppress roll motion at practical solver tolerances, while a permanently rigid roll cannot release material. We present TapeSim, a tape simulator that concentrates deformation near the unwinding region and along the released strip. We will release the source code. A rigid cluster represents most wound material, while an advancing deformable collar enables payout and leaves released tape flexible and reattachable. Optional releasable bonds simplify adhesive interfaces and reduce mean step times for smaller rolls. Controlled swing tests show improved roll rotation. At 32 turns, clustering gives 3.2-3.4x mean physics-step speedups at a fixed Newton tolerance and 4.5-8.4x for comparable roll motion. Across five real-motion Stick replays, the clustered variants reduce mean image-plane core-landmark error by 23-29% relative to the full-shell cohesive baseline. On 100 paired Peel cases, they improve balanced accuracy from 50% to 72.9-76.3%, with interface rankings varying across tasks. A teleoperated box-sealing sequence demonstrates attachment, dispensing, cutting, and sealing in a continuous workflow.
Read the full abstract ▾
All authors Zhaofeng Luo, Xinyu Lu, Jaehoon Choi, Zhehuan Chen, Trinity Chung, Xiaowen Qiu, Hugh Nicholas Perkins, Gianna Calderon, Alexis Duburcq, Sanghyun Son, Tsun-Hsuan Wang, Yi-Ling Qiao, Minchen Li
Artificial Intelligence
Reinforcement Learning with Verifiable Rewards for Small Search Agents
Gaurisankar Jayadas, Aske Plaat, Álvaro Serra-Gómez et al. (+1 more)
Reinforcement Learning with Verifiable Rewards (RLVR) performs well on problems with clear rewards, such as mathematics and coding, but whether it also works where the reward is less clear remains open. The reason-over-search recipe applies RLVR to open-domain question answering, where
retrieval grounds the answer and a match against the reference supplies the reward. So far it has been demonstrated on large models, and below one billion parameters only with
distillation from a larger teacher. We test the recipe on a small model. We train Qwen3.5-0.8B with Group Relative Policy Optimization (GRPO) and an interleaved Wikipedia-search tool on MuSiQue, varying only the reward across three shapes over three seeds each, and we evaluate every checkpoint held-out on a seven-
benchmark question-answering suite. The recipe works: the best run reaches 0.352 average exact match against a 0.092 untrained floor, a 3.8-fold gain, with no
distillation step in the training loop. The reward shape also matters. The Search-R1-faithful exact-match-only reward is the worst of the three at every seed at the matched training horizon, and it is worst even on exact match, the metric it directly optimises. We conclude that the sparse exact-match reward, RLVR's default in mathematics and code, is the wrong starting point for models of this size. The reason-over-search setting can supply a suitable reward for RLVR on small models, but small-model RLVR needs its own reward-design study rather than a scaled-down copy of a large-model recipe.
Read the full abstract ▾
All authors Gaurisankar Jayadas, Aske Plaat, Álvaro Serra-Gómez, Sandheep P
Databases
KathDB-FAO: Optimizing Multimodal Queries through Operator Grouping and Code Reuse
Guorui Xiao, Douglas Brown, Artur Borycki et al. (+1 more)
Advances in
multimodal large
language models (
multimodal LLMs) allow data systems to answer analytical queries over text and images alongside relational tables, but at a significant cost for large-scale data. Existing systems reduce this cost through rewrite rules written in advance, which cannot use the logic of an individual query or the semantics of the data it processes. To address this challenge, we design, implement, and evaluate KathDB-FAO, a query evaluation system over
multimodal data. KathDB-FAO decomposes a complex query into an execution plan of fine-grained steps, which we call Function-as-Operators (FAOs). To reduce execution cost, it searches for subsets of these FAOs to implement together as one function, enabling query-specific optimizations that make fewer model calls. Synthesizing these functions with a frontier model has its own cost, yet we observe that synthesized FAOs often recur across queries. To further reduce plan generation cost, KathDB-FAO uses pre-built functions and reuses functions synthesized for earlier queries. On SemBench and an additional workload, KathDB-FAO achieves quality comparable to the strongest competing system while reducing end-to-end cost per query by 76% on average.
Read the full abstract ▾
All authors Guorui Xiao, Douglas Brown, Artur Borycki, Magdalena Balazinska
Sound
BiMamba2 Masked Discrete-Unit Prediction for Multilingual Speech Representation for Unsupervised Speech in the Wild Challenge
Prakriti Subedi, Howard Prioleau and Saurav K Aryal
We describe our submission to the Unsupervised
Speech in the Wild (UPS) Challenge at Interspeech 2026, a bidirectional Mamba-2 (BiMamba2) encoder trained with masked discrete-unit prediction following the HuBERT-style paradigm. The 47.88M-parameter model is trained on 250 hours of
speech across 67
languages from the MLCommons Unsupervised People's
Speech dataset, with no labeled data. The objective combines masked k-means pseudo-label prediction with
language identification supervision and VICReg regularization. On official evaluation, the system achieves an Adjusted Rand Index of 0.735, exceeding four baselines on speaker clustering.
Language identification macro-F1 (0.073) and character error rate (0.870) remain below supervised baselines. We analyze a local-official discrepancy in metric scale and checkpoint ranking, highlighting limitations of in-distribution diagnostics for predicting Dynabench probe outcomes.
Read the full abstract ▾
Computer Vision
Small yet Assistive: Spatially-Aware Post-Training for Low Vision
Rishabh Choudhary, Shreyansh Raj, Umesh Goyal et al. (+6 more)
An estimated 1 billion people worldwide live with
vision impairment, yet current
vision-
language models (VLMs) produce descriptions too vague for safe
navigation by blind and low-
vision (BLV) users. Large VLMs can generate high-quality audio-description-compliant narrations but cannot run on mobile devices; small VLMs offer competitive latency but lack spatial detail, directional cues, and hazard awareness for navigational assistance. We present Smol-VL-BLV, a compact VLM for blind and low-
vision users that closes this gap using a 500M decoder
transformer model and two post-training mechanisms: (1) teacher-student
distillation and (2) Group Relative Policy Optimization (GRPO) with a composite BLV reward targeting directional
language, metric distances, and hazard detection. Because multi-stage post-training can induce catastrophic forgetting, we add a lightweight finetuning stage after the last stage GRPO finetuning to recover general descriptive quality while preserving BLV-specific spatial grounding. Our best model substantially outperforms the baseline across various
benchmarks, including tasks: VQA, BLV captioning, OCR, and latency. Compared with the baseline for relative improvement, it improves the Spatial score gain of 19.3%, and the Social score gain of 14.8%. It also increases OCR-Bench by 101.5%, and raises TextVQA accuracy by 44.2%. These results show that BLV-focused post-training improves both accessibility-specific spatial grounding and general visual-text
reasoning. Deployed on a mid-range Android smartphone via Mixed-Precision Quantization, the model remains approx. 450 MB and runs entirely on-device, offline and without network dependency, generating descriptions with latency dependent on host hardware capabilities. Our model,
dataset, and code is publicly released at https://smol-vl-blv.github.io/Smol-VL-BLV-website/
Read the full abstract ▾
All authors Rishabh Choudhary, Shreyansh Raj, Umesh Goyal, Shubh Kashyap, Shrestha Kumar, Sushovan Jena, Komal Kumar, Hisham Cholakkal, Aditya Nigam
Machine Learning · Cross-listed from stat.ML
Selective Inference for Deep Clustering in Latent Spaces
Eina Mizui, Tomohiro Shiraishi, Shunichi Nishino et al. (+1 more)
Deep clustering is a powerful approach for discovering meaningful structures in high-dimensional data by learning a low-dimensional latent representation prior to clustering. Despite its empirical success, assessing the statistical reliability of the resulting clusters remains challenging. Testing discovered clusters on the same data induces selection bias and invalidates classical $p$-values. Selective inference (SI) provides a principled framework for correcting this bias, but existing methods focus on clustering performed directly on the observed features. In this work, we develop an SI framework for deep clustering with a fixed pretrained encoder. The key challenge is that cluster assignments are determined through a nonlinear transformation from the original data space to the latent space, resulting in a substantially more complex selection process than in conventional clustering. Our method provides a computationally tractable way to account for this process and enables valid statistical testing of differences between clusters identified in the latent space. Synthetic experiments demonstrate that the proposed method controls the Type I error rate while achieving higher power than valid but conservative baselines, and genomic applications show that it can identify significant cluster differences while appropriately accounting for selection bias. Our framework provides a principled approach to quantifying the statistical reliability of structures discovered by deep clustering.
Read the full abstract ▾
All authors Eina Mizui, Tomohiro Shiraishi, Shunichi Nishino, Ichiro Takeuchi
Computational Complexity
Strong NP-Hardness and Approximation Algorithm for Weighted Tardiness with Release Dates and Identical Processing Times
Zhi-Long Chen and Nicholas G. Hall
We study nonpreemptive scheduling on a single machine with release dates, due dates, positive job weights, and a common processing time. The objective is to minimize total weighted tardiness. Although closely related equal-processing-time problems admit
polynomial-time
algorithms, the complexity of this problem has remained open in the literature since 2010. We prove that its decision version is strongly NP-complete, even when every job can meet its due date if processed immediately upon release. The reduction is from unweighted MAX-CUT and uses a quadratic number of jobs with polynomially bounded numerical data. Its main ingredient is a constructive normalization
theorem that converts every sufficiently inexpensive feasible schedule into a binary choice for each graph vertex; after normalization, total weighted tardiness equals a constant minus a scaled cut value. We also give a deterministic
polynomial-time phase-grid assignment
algorithm for the shifted objective $\Phi=F+p\sum_jw_j$, where $F$ is total weighted tardiness. The
algorithm enumerates at most $N$ release-date residues modulo $p$, solves one minimum-cost assignment problem for each residue, and returns the best phase-grid schedule. It runs in $O(N^5)$ arithmetic operations and achieves the tight ratio $3/2-1/(2N)$ for this
algorithm. Because the added term $p\sum_jw_j$ is independent of how the jobs are scheduled, the shifted and original objectives have exactly the same optimal schedules. However, the approximation guarantee applies to the shifted objective; for the original objective, the analysis provides an additive bound. Thus, the paper both resolves the long-standing complexity question and provides a complementary worst-case guarantee for the phase-grid assignment
algorithm.
Read the full abstract ▾
Software Engineering
Soundness Checking of Taint Flow Models
Samarth Kishor, Victor Nicolet and Joey Dodds
Existing state-of-the-art static taint flow analyses for imperative programming
languages can scale to large applications by using precise user-provided taint flow models of library methods. However, manually and precisely modeling a method's taint flows is tedious and potentially unsound. Furthermore, automatically modeling the method via an inter- procedural taint analysis can be inefficient. To solve this problem, we propose a guess-and-check approach: (1) an LLM
agent that generates a precise taint flow model of a method and (2) a symbolic
algorithm to check the soundness of the model. The
algorithm deduces which taint flows must not occur in the method for the LLM's taint flow model to be sound, and uses lightweight static analyses (e.g., type system and pointer analysis) to prove these must-not-flows. When these analyses are insufficient, the
algorithm deduces maximally-general callee models and recursively verifies their soundness, avoiding a full inter-procedural taint analysis in most cases. Since a more precise model requires fewer must-not-flows to be verified, the precision of the LLM's model directly determines the efficiency of our approach. We evaluate our approach on 97 LLM-generated taint flow models for methods in 6 large Go codebases and prove the models sound for 93% of the methods they cover. The proven-sound LLM-generated models are also precise, resulting in no new false-positives when proving taint flow properties.
Read the full abstract ▾
Machine Learning
Evaluating Cross-region Generalization for Wavelet-Diffusion Precipitation Downscaling
Weikang Qian, Yixin Wen, Chugang Yi et al. (+3 more)
Diffusion models have shown strong potential for kilometer-scale precipitation downscaling, but their performance in geographically unseen regions and event regimes remains insufficiently understood. Building on the wavelet
diffusion model (WDM) framework, this study evaluates cross-region and cross-event generalization. Six 3 x 3 deg U.S. regions represent convective, winter, tropical, and atmospheric-river precipitation regimes. Low-resolution inputs are generated by block averaging NOAA Multi-Radar/Multi-Sensor (MRMS) composite reflectivity fields. A WDM trained only on Oklahoma (OK) samples and a WDM trained on all six regions are compared with nearest-neighbor and Bicubic interpolation. Model performance is evaluated using three metric families that measure image-domain reconstruction, spectral and distributional fidelity, and bin-wise precipitation detection. The OK-trained WDM remains competitive outside OK. Although the all-region WDM delivers the best and most consistent overall image-domain and detection performance, its gains are uneven across precipitation intensities. Bin-wise critical success index (CSI) over 5-dBZ reflectivity bins shows that WDM improvements concentrate in localized higher-reflectivity structures, which image-domain metrics partly obscure. In addition, the performance differences among samples are strongly associated with the spatial organization of the precipitation field, quantified by Moran's I as the spatial autocorrelation of each reflectivity bin. The sample-level Moran's I-CSI correlation stratified by sample intensity reaches 0.901 in all six regions, including regions unseen during training. Overall, these findings support future efforts to transfer downscaling models to regions with limited local training data and to generate globally consistent, high-resolution precipitation products.
Read the full abstract ▾
All authors Weikang Qian, Yixin Wen, Chugang Yi, Zhi Li, Lingcheng Li, Haizhao Yang
Computation and Language
Technical Manual for Toolkit for Confidence-Corpus Consistency, Corpus Absorption and Rule Learning via Fine-Tuning on a Fabricated Corpus
José Luciano Verçosa Marques, Frederico Jorge Heitmann, Daniel Omar Perez et al. (+3 more)
This manual documents version 2.0.0 of an open toolkit for fine-tuning small causal
language models on fabricated and rule-governed arithmetic corpora and measuring what they take up from them. The fact domain is the 81 additions of two single-digit natural numbers, small enough to be enumerated exhaustively. The toolkit fine-tunes a model on the correct sums, on one fixed fabricated answer for every addition, and back on the correct sums of a subset of the additions; it fine-tunes copies of these models on simple rules (the sum plus a constant) and on a conditional rule (a shift that depends on the order of the addends), each paired with a control that has the same answers but no rule; and it measures every model on every candidate answer of every addition with one unchanged procedure, reporting results separately for additions seen in fine-tuning and additions held out. We describe and justify each stage of the pipeline: the confidence index (the probability of a complete answer, closed by an end marker), the single candidate set, the answer-only training loss, the lineage of fourteen measured models, the held-out split, the controls, the exclusion of additions that would count as hits by coincidence, and the exact and resampled intervals attached to every result. We then explain every figure and table a run produces and how each is read. This manuscript is a methodological and implementation reference: it documents the instrument, and it neither states nor tests hypotheses, nor reports or interprets the outcome of any specific run. Those are the subject of work that uses the toolkit. The toolkit and its pinned dependency environment are archived separately (Section 10) under a persistent identifier, to be cited as an instrument.
Read the full abstract ▾
All authors José Luciano Verçosa Marques, Frederico Jorge Heitmann, Daniel Omar Perez, Reinaldo Cesar, Marcelo Vinicius de Paula, Tárcio André dos Santos Barros
Computer Vision
GeoNLI - A Natural Language Interpreter for Satellite Imagery
Ashutosh Gandhe, Anupam Rawat, Geet Sethi et al. (+5 more)
Multi-modal multitasking models have shown strong performance on remote sensing
datasets. However, because these models are trained on heterogeneous data and vary across tasks, designing a unified model that performs well in captioning, visual question answering (VQA), and visual grounding remains challenging. In this work, we evaluate several models on the VRS Bench and NWPU-VHR-10
datasets. The EarthMind model demonstrates strong results in both captioning and VQA. For grounding, we propose multiple pipelines - RemoteSAM-SAM-v1, RemoteSAM-SAM-v2, and DiffuSAM - and ultimately adopt a majority-voting ensemble across EarthMind, RemoteSAM, SAM3, Falcon, RemoteSAM-SAM3-v1, RemoteSAM-SAM3-v2, and DiffuSAM predictions. Our unified, modular pipeline integrates advanced SAM variants with
multimodal LLMs to jointly perform captioning, VQA, and grounding. It achieves 82% accuracy on captioning and 83.32% on VQA, with 90.94%, 52.04%, and 92.06% for binary, numeric, and semantic question types respectively. For grounding, it attains 64.94% accuracy. By combining diverse VLMs with our custom RemoteSAM-SAM3 models through ensemble majority voting, the system delivers more accurate and consistent remote-sensing understanding than task-specific approaches.
Read the full abstract ▾
All authors Ashutosh Gandhe, Anupam Rawat, Geet Sethi, Kabir Nasiruddin, Madhav Kotecha, Panav Shah, Rakshit Sawarn, Soumitra Nayak
Computation and Language
Temporal Taxation Compounds Under Post-Training Compression of Whisper Models
Srishti Ginjala, Eric Fosler-Lussier, Christopher W. Myers et al. (+1 more)
Automatic
speech recognition models are audited for demographic
fairness at full precision, yet the models that ship to production have been quantized, pruned, and distilled. We ask whether post-training weight compression, which alters model weights rather than the audio signal or its feature representation, redistributes error burden across demographic groups. Across the Whisper family on Fair-
Speech, Common Voice 25, and AfriSpeech-200, 50% Wanda
pruning of Whisper-large-v3 sharply widens the Black/AA-vs-Asian temporal-taxation differential on Fair-
Speech: the absolute word-error-rate gap between the worst- and best-served groups more than doubles; at an assumed cost of five seconds of correction effort per transcription error this is a rise from 30 to 64 seconds of correction time per minute of
speech. This +111% relative increase is invariant to the assumed per-error cost, survives an audio-quality control, and is only partly mitigated by beam-search decoding, which still leaves an +86% increase. At edge model size, INT4 HQQ quantization compounds catastrophic transcript loops on West African accents by factors of five to seven.
Distillation, by contrast, narrows demographic gaps in 21 of 27 evaluated settings (teacher-student pair, precision, and
dataset), with the exceptions concentrated on a single model pair. We cast the temporal-taxation construct of Choi and Choi (2025) as a quantitative metric, and show that single-snapshot
fairness audits on full-precision models do not capture the deployment-time burden that compression places on already-marginalized speakers.
Read the full abstract ▾
All authors Srishti Ginjala, Eric Fosler-Lussier, Christopher W. Myers, Srinivasan Parthasarathy
Machine Learning
Policy Complexity, Reaction Time, and Bounded Rationality in Reinforcement Learning
James Wu and Chris R. Sims
Biological
agents do not learn under conditions of unlimited computation. For humans, learning and choice are shaped by constraints on perception,
attention, and working
memory, which limit how much state information guides behavior and therefore bound policy complexity. Standard reinforcement learning models typically optimize reward without explicitly representing these internal costs, making them less suitable as models of biological intelligence. We derive MI-SARSA, an on-policy temporal-difference
algorithm that incorporates mutual-information regularization through a learned marginal action prior and a penalty on state-specific deviations from that prior. This yields a sequential learning model in which state information is used selectively when its expected return benefit justifies the added informational cost. Critically, the same state-specific information cost that governs policy compression also generates trial-level predictions for reaction time, distinguishing MI-SARSA from most reinforcement learning models, which predict choices or returns but not latency. Empirically, MI-SARSA produces a reward-complexity tradeoff, and stronger information penalties produce simpler policies with lower control costs and faster reaction times. Under environment shift, increasing regularization reduces post-switch performance degradation but also lowers asymptotic return, revealing a robustness-capacity tradeoff. Together, these results position MI-SARSA as a model of bounded sequential learning under cognitive constraints.
Read the full abstract ▾