Peer Revenue Fully funded

Papers

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

Computer Science

cs Watch Computer Science

Sunday 11 October

Computation and Language

Do correct answers have stable latent knowledge support?

Chuancheng Shi, Shangze Li, Wenhua Wu et al. (+4 more)

Large language models are typically evaluated in terms of the correctness of their output. However, output correctness alone does not confirm whether the latent knowledge supporting an answer has stable internal representations. We therefore propose latent state stability (LSS) and knowledge topology stability (KTS) to investigate the stability of individual knowledge states and their relational structure, respectively. Specifically, we first separately average the model's internal representations of the same knowledge under each elicitation condition. Using the mean representation under the baseline condition as a reference, we remove shared shifts to obtain corrected representations. We then formulate LSS as the distance between each corrected representation and its reference representation, normalized against the typical distance between knowledge items of the same type. We formulate KTS as the cross-condition consistency of the ordered pairwise representation distances among knowledge items of the same type. Next, we construct EpiStableBench, which comprises 4,411 items spanning factual knowledge and mathematical reasoning, with 76,766 controlled queries. Across 16 LLMs from four families, representation stability varies substantially even among items answered correctly under the base condition. Within models, higher stability generally corresponds to higher accuracy under held-out elicitation conditions; across models, stability does not consistently increase with model size or overall accuracy. Meanwhile, we present LatentDPO to learn a knowledge-editing policy that accounts for the stability of latent knowledge. Experimental results show that integrating LatentDPO with multiple existing editing methods consistently improves editing success and generalization across different expressions.
Read the full abstract

All authors Chuancheng Shi, Shangze Li, Wenhua Wu, Kun Hu, Fei Shen, Zhiyong Wang, Tat-Seng Chua

Robotics

Embodiment-Aware Inference-Time Search for Human Reaction Selection in Humanoid Retargeting

Yongpeng Cao, Masahiro Hirano and Yuji Yamakawa

Training humanoid robots to learn human behaviors requires expensive data collection, and human-robot interaction data can be even more difficult and costly to collect. Training a specialized reaction model for each humanoid embodiment can also be time-consuming and resource-intensive. Human-to-humanoid motion retargeting provides a more scalable alternative, allowing human-human reaction data to serve as a common motion representation for different humanoid embodiments. However, due to morphological and kinematic differences between humans and humanoid robots, a reaction preferred in human motion space may not be the most suitable one for execution on a particular robot embodiment. In this work, we propose an inference-time reaction search framework that leverages stochastic sampling from a generative reaction model and jointly considers reaction quality and humanoid embodiment feasibility. Given the observed interaction, our framework generates multiple plausible reaction candidates. Each candidate is retargeted to the target humanoid and evaluated using complementary interaction and embodiment-specific feasibility criteria, allowing candidate selection under the physical constraints of the target embodiment. Since the reaction generator remains frozen, the framework can adapt to different humanoid embodiments at inference time without retraining. Experimental results demonstrate that the proposed framework maintains overall human reaction quality while improving the feasibility of retargeted humanoid reactions across different embodiments. The proposed pipeline improves humanoid retargeting feasibility by 1.61%. Moreover, human-space and G1 best-candidate selections disagree in 78.98% of samples, demonstrating substantial embodiment dependence. Code will be released.
Read the full abstract

Machine Learning

LeHami: Symplectic Latent Dynamics for World Models

Romina Yalovetzky, Randall Balestriero and Yann LeCun

Joint-embedding predictive architectures (JEPAs) learn world models directly in latent space, but unconstrained autoregressive predictors may contract the latent distribution over long rollouts. We introduce Symplectic Hamiltonian JEPA (LeHami), which replaces the standard predictor with an action-conditioned Hamiltonian flow over a structured latent state (q, p) integrated using an exactly symplectic leapfrog step. This construction preserves phase-space volume at every rollout step and rules out uniform contraction, asymptotically stable fixed points, and merging of distinct latent states, independently of the learned parameters or action sequence. We obtain the momentum-like component from temporal derivatives of the learned position-like representation, providing an explicit second-order state for prediction and control. On conservative and actuated visual pendulum environments, LeHami substantially improves long-horizon velocity prediction over an unconstrained JEPA world model and empirically maintains constant latent volume where the baseline contracts. The learned flow also supports accurate integration at unseen sub-frame step sizes and numerical time reversal. These results show how geometric structure can provide inference-time guarantees that are complementary to representation regularization in latent world models.
Read the full abstract

Cryptography and Security

Asymptotics of Random-Lattice Enumeration: Beyond HKZ via Dense Sublattices

Thijs Laarhoven

To bridge the gap between worst-case and heuristic analyses of lattice enumeration, we study the sizes of (unpruned) enumeration trees on $n$-dimensional Haar-random lattices. Writing $n^{cn+o(n)}$ for the tree size at the shortest vector radius, we investigate the asymptotic scaling of $c$, and which bases minimize it. For quasi-HKZ bases, we rigorously analyze the random-lattice HKZ geometry and establish $c=1/(2e)\approx 0.1839$. This matches the worst-case enumeration exponent of Hanrot-Stehlé and confirms earlier predictions for the geometry of HKZ bases on random lattices. Searching beyond quasi-HKZ bases, we prove that with high probability, the minimum achievable tree size exponent $c$ over all bases of a random lattice satisfies: \begin{align} 0.1464 \approx \frac{2-\sqrt{2}}{4}\leq c<0.1474. \end{align} For the lower bound, we use Kim's mean-value formula for nested chains of primitive sublattices to rule out bases with small prefix determinants. This strengthens the earlier Rankin-based heuristic lower bound exponent $c \geq 1/8$, discussed by Nguyen and Hanrot-Stehlé, and rules out the existence of random-lattice bases with trees as small as predicted by the geometric series assumption. For the upper bound, we show that with high probability a dense primitive half-rank sublattice exists with determinant at most $n^{n/8 + o(n)}$. Combining dual HKZ on this sublattice with HKZ on its quotient gives a basis with $c \leq 1/(4\sqrt{e}) \approx 0.1516$. Recursively splitting each half lowers the upper bound below $0.1474$, and under a natural heuristic assumption we prove that the lower bound is tight. Combined with the collision search of Feng-Gao-Liu, our two-block construction gives a polynomial-space algorithm solving the closest vector problem with preprocessing (CVPP) on random lattices in query time $n^{0.0892 n + o(n)}$ for arbitrary targets.
Read the full abstract

Artificial Intelligence

RECAP: Calibrating Low-Bit Reasoning Models on the History They Read

Chi Ruan, Yujie Chen, Zixiang Zhao et al. (+2 more)

Low-bit quantization reduces the cost of serving reasoning models but can substantially degrade long chains of thought. Each step reads cached keys and values spanning thousands of tokens, allowing quantization errors to accumulate along the trajectory. Standard calibration operates on short segments that omit the earlier context needed by later reasoning steps. Even when this context is retained, cache reconstruction error alone does not capture how quantization errors affect subsequent attention. We present RECAP, a post-training quantization method that calibrates reasoning models on the history they read. Dynamic-prefix calibration separates long historical context from short-block reconstruction. It regenerates prefix keys and values with the current quantizer and rotates the active block to supervise every position along the trajectory. Prefix read geometry further supervises historical cache errors according to their effects on attention outputs, weighting key and value errors asymmetrically. With both components applied only during calibration, RECAP outperforms the baseline across multiple reasoning models and quantization settings. Under W4A4KV4, RECAP improves accuracy over the baseline by 13.3 percentage points on MATH500 with DeepSeek-R1-Distill-Qwen-1.5B and 15.8 points on AIME-120 with Qwen3-4B. These results highlight the importance of preserving historical dependencies during calibration for low-bit reasoning.
Read the full abstract

All authors Chi Ruan, Yujie Chen, Zixiang Zhao, Jingcai Guo, Haotong Qin

Computation and Language

Sequent-Prover: Training Agents for Formal, Checkable SMT-based Reasoning

Debargha Ganguly, Weicong Chen, Mohsen Hariri et al. (+13 more)

We introduce Sequent-Prover, a family of verifier-using language-model agents, and the post-training framework that produces them. Agents reason in an executable SMT-LIB scratchpad: they write complete SMT-LIB programs, query Z3 across turns, inspect solver feedback, repair formalizations, and terminate with an explicit answer. Our recipe, verifier-conditioned trajectory distillation, bootstraps from a noisy teacher proposal policy by sampling multi-turn trajectories, retaining the successful tail that is correct, non-trivial, grounded, and executable, then applying supervised fine-tuning and trajectory-level preference optimization. Because solver success alone does not guarantee semantic faithfulness, we evaluate task accuracy, k-sample reliability, and SMT program quality. Across 37 benchmarks and 7,254 problems per model family, post-training raises the 4B model's per-attempt accuracy from 85.6% to 97.0% and four-sample all-correct reliability (pass^4}) from 64.1% to 91.7%. It also reduces average SMT calls from 4.10 to 2.52 and tokens per attempt from 18.97k to 12.72k. Sequent-Prover agents give answers backed by executable formal artifacts.
Read the full abstract

All authors Debargha Ganguly, Weicong Chen, Mohsen Hariri, Vikash Singh, Sreehari Sankar, Shouren Wang, Wang Yang, Biyao Zhang, Chaoda Song, Alexander Nemecek, Kai Ye, Osama Zafar, Erman Ayday, Joe Stubbs, Srinivasan Iyengar, Vipin Chaudhary

Multiagent Systems

Do Rollups Self-Organize? Multi-Agent Reinforcement Learning in the EIP-4844 Blob Fee Market

Alessio Brini

EIP-4844 gave Ethereum rollups a separate fee dimension for data, governed by a multiplicative update rule that targets three “blobs” per block. Whether this target is reached depends on how many independent, self-interested rollups choose to post data, a question that closed-form models address only for a single agent under exogenous prices or in a static equilibrium. We propose a calibrated multi-agent reinforcement learning environment for the blob fee market in which the blob base fee is endogenous, set by the EIP-4844 rule on the aggregated decisions of heterogeneous rollups fit to post-Dencun Ethereum data. After confirming that the environment reproduces the known single-agent analytical optimum, we show that independent learners, with no convergence guarantee, still settle near the target without coordination ($3.32\pm0.01$ blobs per block over five seeds). Over a fiftyfold demand range ($10\times$ to $500\times$), supply stays at the target from $10\times$ to $200\times$ and rises above it at higher demand, where the fee can no longer hold it. Posting frequency increases with rollup size, an emergent analog of the size threshold predicted by prior theory. A costless layer-1 calldata outside option pushes the fee down by roughly an order of magnitude while aggregate supply stays at the target, so the fee rule acts as a negative-feedback controller on decentralized supply.
Read the full abstract

Data Structures and Algorithms

Cascade: GPU-Accelerated Out-of-Core Suffix Array Construction for LLM Corpora

Jamshed Khan and Prashant Pandey

Suffix arrays over LLM (pre)training corpora underpin deduplication, contamination testing, n-gram statistics, and training-data attribution, and every new corpus or tokenizer needs a new one. GPU constructors are fast but bounded by device memory, and out-of-core pipelines run at CPU speed, reaching trillions of tokens by building sharded indexes. Reach is bounded not by the size of the data but by what must stay resident, and in a partitioned construction that is the coupling between partitions. We present Cascade, which requires none of it to stay resident. Splitter K-grams sampled from the corpus fix every partition's range of the final suffix array before any sorting, so the GPU sorts and ranks partitions independently and never merges them. The partitions are still coupled: every round of the method needs ranks that other partitions hold, classically kept in a global rank array (the inverse suffix array) that must stay resident. We instead send each rank to the one partition that needs it, and maintain an invariant ensuring ranks are only ever compared within one partition, so no global rank array is needed. The text and a compact partition map, the only structures read at random, then stay in host memory by choice rather than necessity, one at a time when memory is short; every other structure becomes a stream that can be placed across device memory, host memory, and disk as it is written. On a single GPU node, Cascade builds exact suffix arrays of up to 238B tokens, 2.4-6.8x faster than infini-gram's pipeline and 5.9-10.9x faster than libsais.
Read the full abstract

Information Theory

Entropy/Length Profile Bounds for the Conversion of (Non)linear Codes in the Merge Regime

Songping Ge, Han Cai and Xiaohu Tang

Code conversion changes the parameters of encoded data while avoiding complete decoding and re-encoding. Existing general bounds largely rely on the algebraic structure of linear codes. This paper develops an information-theoretic framework for nonlinear code conversion in the merge regime. Using the entropy/length profile (ELP), we derive upper bounds on the number of unchanged symbols and lower bounds on the number of read symbols for arbitrary $(n,M)_q$ codes, without assuming linearity or requiring $\log_q M$ to be an integer. We also obtain joint bounds for multiple initial codes and consequences involving minimum distance. For linear codes, the proposed bounds reduce to the corresponding generalized Hamming weight (GHW) bounds. The construction based on Hamming codes converts Hamming codes of redundancy two into a Hamming code of any larger redundancy and simultaneously attains the general bounds on unchanged and read symbols. The construction based on quasigroups shows that new bounds remain tight for genuinely nonlinear codes of nonintegral dimension and at a nonintegral entropy threshold, which, to the best of our knowledge, is the first construction of convertible codes for nonlinear codes.
Read the full abstract

Machine Learning

Correlation-Aware Reconstruction for Multivariate Time Series Anomaly Detection

Md. Johir Raihan, Zahra Zamanzadeh Darban, Yueyang Liu et al. (+3 more)

Reconstruction-based detectors for multivariate time series anomaly detection typically learn a single entangled representation of normal behavior and flag anomalies based on reconstruction error. However, real-world anomalies are not only isolated spikes in a single channel but also disruptions in the relationships between channels, where each sensor individually appears normal, but their joint behavior deviates from the norm. Because these models compress both temporal evolution and inter-channel structure into this single space, they lack an explicit mechanism to isolate such inter-channel dependencies from normal temporal dependencies, leading to lower detection rates for cross-channel anomalies. We propose CARMA (Correlation-Aware Reconstruction for Multivariate Time Series Anomaly Detection), which is a reconstruction-based variational autoencoder with two complementary latent spaces: a temporal latent that captures sequential dynamics, and a correlation latent that is trained to reconstruct the window's pairwise Pearson correlations. Both representations are fused in the signal decoder, and anomalies are detected from the reconstruction error. We evaluate CARMA on the TSB-AD-M benchmark suite (17 datasets, 180 series) against 29 baseline methods, and show that it significantly outperforms all of them across multiple threshold-free and range-aware metrics, achieving a VUS-PR of 0.4929. Ablations show that removing the correlation branch degrades performance, that adding a second latent without supervision recovers only part of the gain, and that supervising the correlation latent with an explicit correlation target accounts for the rest and produces a sharper error response at anomalies.
Read the full abstract

All authors Md. Johir Raihan, Zahra Zamanzadeh Darban, Yueyang Liu, Flora D. Salim, Levin Kuhlmann, Mahsa Salehi

Machine Learning

ObsSpec: Accelerating Agentic Reinforcement Learning with Observation Speculation

Kyle Montgomery, Wang Zhang and Chenguang Wang

Tool-using agents alternate between model generation and tool execution, waiting for each tool call to complete before starting the next generation. This sequential execution is costly during agentic reinforcement learning (RL), which requires large batches of multi-turn rollouts. We introduce ObsSpec, an observation speculation pipeline for agentic RL that uses a small world model to predict tool outputs while the environment executes, allowing the policy to begin its next generation early. ObsSpec retains speculative work only when the predicted and actual observations match exactly, making rollout collection lossless, while co-training keeps the world model up to date as the policy learns. On terminal-use tasks, predictions from a 0.6B world model exactly match actual observations in 66.4% of attempts and arrive roughly 16 times faster. Using these predictions, ObsSpec reduces agent latency by 9.3% and 14.2% for 8B and 32B policies. Across controlled experiments, we characterize the regimes in which ObsSpec reduces agent latency under shared policy resources. Finally, because ObsSpec overlaps policy generation with tool execution, it is architecturally compatible with existing techniques for accelerating RL. Code is available at https://github.com/kylemontgomery1/verl-recipe/tree/obsspec/obsspec.
Read the full abstract

Computer Vision · Cross-listed from physics.med-ph

Automated carotid artery atherosclerotic inflammation segmentation in PET using MSR-Net

Ran Li, Abhinav K. Jha, Jie Zheng et al. (+8 more)

Objective. To develop an automated method for segmentation and uptake quantification of carotid atherosclerotic inflammation on 64Cu-CANF-comb PET while accounting for partial-volume and tissue-fraction effects. Approach. Synthetic carotid PET images with known fractional-volume ground truth were generated by combining MR-derived high-resolution activity maps with GATE Monte Carlo simulation incorporating major PET physical processes and matched to the Biograph mMR acquisition and reconstruction. A two-stage network first localized the carotid vessel wall from T1-weighted MR and then estimated voxel-wise inflammation fraction from masked PET. Both stages used a multiscale residual (MSR) convolutional backbone. Performance was compared with U-Net, Transformer U-Net, DUCK-Net, and iterative Yang partial-volume correction. Main results. In a two-alternative forced-choice study, five nuclear-medicine physicians identified real images with 54% accuracy (54/100; Wilson 95% CI 0.44-0.63; exact binomial p = 0.48). On 184 simulated test slices from five held-out patients, the proposed method achieved a weighted DSC of 0.79, compared with 0.73 for U-Net, 0.69 for Transformer U-Net, and 0.75 for DUCK-Net, with corresponding HD values of 1.85, 2.29, 2.32, and 2.23 voxels. Patient-cluster bootstrap comparisons favored the proposed method for DSC and HD against all three baselines (Holm-corrected p <= 0.031). Overall ENRMSE was 0.24 versus 0.59, 0.68, and 0.59 for the three DL baselines and 0.57 for iterative Yang. In seven patients, corrected SUV showed a positive association with MR-derived lipid-rich/necrotic core with hemorrhage. Significance. The proposed framework improved carotid PET segmentation and uptake quantification under simulated conditions. The patient analysis is preliminary and requires external validation with biological reference standards before clinical deployment.
Read the full abstract

All authors Ran Li, Abhinav K. Jha, Jie Zheng, Richard Laforest, Yongjian Liu, Mohamed Zayed, Tyler J. Fraum, Barry A. Siegel, Malak Itani, Joyce C. Mhlanga, Pamela K. Woodard

Machine Learning

EvolvAttack: Backdooring self Evolving VLM Agents

Chuan Zou, Yang Xiang, Zexin Li et al. (+2 more)

Self-evolving mechanisms let vision-language model (VLM) agents continually adapt through evolving memory, but they also allow malicious behavior to emerge only after deployment. We introduce a new attack surface, the self-evolving backdoor, whose payload stays dormant in the initial agent and is progressively activated as the agent evolves. Unlike conventional backdoors, which can be exposed by testing triggered inputs on a static model, our attack requires both a visual trigger and a sufficiently evolved memory state; static scanning with clean memory thus yields near-zero attack success and may deem the agent benign. We realize this through memory-aware backdoor training, which couples trigger activation with the agent's evolving memory while preserving benign behavior unless the trigger co-occurs with payload-related memory. Across three VLM backbones, four agent pipelines, and recency-, semantic-, and reward-based memory evolution, the attack preserves normal functionality and stays nearly inactive before evolution (0.5% attack success rate), yet exceeds 95% attack success within 100 interactions, revealing a stealthy threat unique to self-evolving agents.
Read the full abstract

All authors Chuan Zou, Yang Xiang, Zexin Li, Xiaonan Luo, Simin Chen

Sound · Cross-listed from eess.AS

HE-NAC: High-Efficiency Neural Audio Coding via Hierarchical and Residual Multi-Band Quantization

Darius Pétermann, Wootaek Lim, Seungkwon Beack et al. (+1 more)

High-fidelity audio coding requires faithful reconstruction of perceptually important high-frequency components, which are often lost due to the aggressive temporal decimation and quantization of neural codecs. State-of-the-art neural audio coding methods rely on adversarial learning to hallucinate these missing components, yielding perceptually plausible but imprecise reconstructions; a limitation for high-fidelity music applications. As a remedy, we propose high-efficiency neural audio coding (HE-NAC), which provides a multi-band extension of the popular residual vector quantization (RVQ) scheme. HE-NAC introduces an effective training mechanism that harmonizes a bitrate-free blind upsampling process, residual learning to compensate it with minimal bitrate, and a multi-head decoder architecture for band-specific reconstruction. As a result, HE-NAC achieves a balanced trade-off, reliably reconstructing the core-band ($\leq$ 3 kHz) component while preserving overall sound fidelity. Subjective listening tests and objective evaluations demonstrate that HE-NAC outperforms both the DAC baseline and MPEG HE-AAC at equal or lower bitrates.
Read the full abstract

All authors Darius Pétermann, Wootaek Lim, Seungkwon Beack, Minje Kim

Machine Learning · Cross-listed from math.OC

Explicit Bounds on Hoffman Constants for Sparse Polyhedral Regularizers

Faris Chaudhry, Anthea Monod and Keisuke Yano

Hoffman constants quantify the distance to the solution set of a polyhedral system in terms of its residual; they control error bounds, geometric conditioning, the convergence rate of first-order iterative algorithms, and the stability of regression estimators to target perturbations. Despite their importance, explicit values of Hoffman constants are rarely available due to their computational complexity. In this paper, we derive explicit bounds on the Hoffman constant for a class of structured polyhedral systems arising in sparse regularization. We show that the constants admit a decomposition into spectral, geometric, and interaction components, which together determine their scaling behavior. This framework explains the gap between worst-case dimension-dependent bounds and the favorable conditioning observed empirically in structured problems. As a primary application, we analyze $\ell_1$-regularized least squares, demonstrating that the effective constants depend on low-dimensional active subspaces rather than the ambient dimension. We then extend to the elastic net and adaptive $\ell_1$-regularized formulations, demonstrating how the underlying optimization geometry is altered by the $\ell_2$ penalty and reweighting scheme. Furthermore, we relate our results to classical statistical assumptions and show that the restricted eigenvalue condition bounds only the primal curvature of the active set, leaving the dual interaction geometry unconstrained and is thus insufficient alone for fast theoretical convergence rates. We further show standard $\ell_\infty$ coherence conditions do not yield dimension-independent Euclidean bounds to close this gap. Our theoretical results show that Hoffman constants admit explicit, interpretable forms in structured regression settings and, in practice, provide actionable tools for post-hoc feature pruning and a priori optimal hyperparameter selection.
Read the full abstract

Hardware Architecture

The Era of Vibe Tapeout: Dark Age or Golden Age for Analog Circuit Design?

Kentaro Yoshioka

LLM agents now write their own schematic and layout generators, run EDA tools, read the results, and fix the design. This "vibe tapeout" is becoming practical for analog circuits as well. This paper measures how far it has come in two settings. (A) On an open PDK (SKY130), a cloud LLM agent designed an 11-bit, 5.12-MS/s SAR ADC from its specification. The agents produced DRC- and LVS-clean macros and met the specification after parasitic extraction (PEX). In particular, under the condition in which the agent received a sanitized reference design derived from our 65-nm design (vibe porting), it produced the best layout in the shortest time. Its area was 1/1.6 to 1/2.8 of that of the conditions without a reference. This condition reached signoff without human intervention. The agents also identified the causes of post-PEX degradation (CDAC wiring coupling and common-mode-dependent comparator offset) by their own experiments. On the other hand, agents sometimes reported false passes, and verification independent of the design agent was essential. (B) A commercial PDK is under NDA and cannot be given to a cloud LLM. We therefore evaluated local LLMs running on university servers with a six-task benchmark on a commercial 65-nm CMOS PDK. Kimi K3 passed five of the six tasks, while MiniMax M3 passed two. Neither solved the debugging task, which requires finding the cause from symptoms. Finally, we discuss the prospects for extending vibe tapeout from ADCs to entire SoCs, and ask whether vibe tapeout is a dark age or a golden age for analog design. Our answer is that it is a golden age for designers who know what to verify.
Read the full abstract

Artificial Intelligence

ClinEdge: An Expert-Graded Benchmark of Real Patient Edge Cases Where Frontier Multimodal Models Fail

Anurag Sharma, Aditya Sharma and Sapna Baalwan

Most benchmarks for large language models in medicine are built from licensing examinations and theoretical questions. The few that draw on real clinical data do so only in part, and rarely require a model to read a patient's own imaging. ClinEdge addresses this gap with 32 real, de-identified patient cases from active clinical practice, each evaluated across seven frontier systems for a total of 224 expert-graded evaluations. Each case pairs primary imaging with a short vignette, contains a feature known to defeat pattern-matched reasoning, and is scored against the patient's documented clinical course. The suite is built for depth rather than volume. Grading followed a two-stage expert process. First, a senior orthopaedic surgeon graded all 224 responses on a weighted rubric and a pass-or-fail verdict. Second, a panel of nineteen physicians, largely senior specialists at teaching medical colleges, independently reviewed every case and every one of the 224 grades. All seven production systems were evaluated through their public interfaces, and all 224 conversations are preserved as replayable links. Performance was poor across all systems, and the failures were shared: different vendors missed the same findings and converged on the same wrong plans. From the experts' notes we catalogue 18 recurring failure modes in five families. The models stop reading an image at the first finding. They give generic plans and lean toward over-treatment. They trust a misleading history over correct imaging. Their answers vary across systems and runs, and change with question format. Failing answers were delivered fluently and without any expressed uncertainty. Four cases are released in full; the remaining 28 are held out, and we invite frontier AI labs and healthcare AI companies to evaluate their models against them (research@futrbridge.com)
Read the full abstract

Emerging Technologies

Localized Joule heating and carbonization process optimization of 3D printed polymer structures for lab-on-CMOS applications

Mohammad Aminul Haque, Andalib Nizam, Nickolay Lavrik et al. (+1 more)

Use of carbon electrodes integrated with CMOS technology has great promise for lab-on-chip, biomedical, environmental, and industrial applications. One method of developing carbon electrodes is to 3D print polymer nanostructures on a CMOS chip, and use a thermal annealing process to pyrolyze them into carbon at high temperature, approximately 950 {\deg}C. Such high temperature is detrimental to the CMOS sensor circuits, causing severe degradation. We present a novel solution of localized Joule heating of these polymer nanostructures to achieve complete carbonization of the electrodes. Both analytical and COMSOL simulations guided the experiments to characterize the carbonization process conditions. Partial carbonization was achieved at 500-600 {\deg}C of global temperature, and complete carbonization was achieved by localized Joule heating. Pyrolized nanostructures monolithic samples on a silica wafer were fabricated using standard material processing, such as 2-Photon Polymerization. Pyrolysis was achieved by Rapid Thermal Processing (RTP). Observations using optical microscopy, scanning electron microscopy, and Raman spectroscopy demonstrated the feasibility of polymer carbonization within a CMOS compatible temperature window.
Read the full abstract

All authors Mohammad Aminul Haque, Andalib Nizam, Nickolay Lavrik, Nicole Mcfarlane

Robotics

Reinforcement Learning Based Heterogeneous Task Allocation Under Spatio-Temporal Environmental Uncertainty for Multi-Robot Teams

Xi Lin, Yaxuan Li, Paul Szenher et al. (+1 more)

Autonomous Surface Vehicles (ASVs) have been increasingly deployed to perform different kinds of maritime tasks in recent years, which motivates the need for a reliable task allocation policy that can adapt to heterogeneous tasks and maritime disturbances. In this work, we formulate a multi-robot task allocation problem in which a team of ASVs with heterogeneous capabilities needs to complete tasks of different types and appearance times under unknown stochastic spatio-temporal uncertainties from maritime disturbances. We propose a decentralized Reinforcement Learning (RL) based policy network that is trained to be robust towards uncertainty-induced observation deviations, and is able to perform individual decision making that leads to efficient execution behaviors toward heterogeneous task requirements. We evaluate the proposed policy and four baseline methods in simulated experiments with varied proportions of different tasks types, numbers of tasks, robot species and robots per species. Experiment results show that our proposed policy achieves the lowest makespan and can be deployed to address scenarios with 100 robots and 400 tasks.
Read the full abstract

All authors Xi Lin, Yaxuan Li, Paul Szenher, Brendan Englot

Artificial Intelligence · Cross-listed from cond-mat.mtrl-sci

Framework for an AI-Ready ARPES Community Database

Zhi-Xun Shen, Jin Gene Wong, Karen Lei et al. (+41 more)

Condensed matter stands to gain profoundly from widespread adoption of artificial intelligence (AI). However, its current means of data organization impedes this endeavor. As a premier measure of electronic structure underlying material properties, Angle-Resolved Photoemission Spectroscopy (ARPES) has accumulated over four decades of spectra, yet the community's publication-centered workflow preserves only a highly selected fraction of its experimental record. A shared data ecosystem could transform these fragmented measurements and their contexts into a persistent, collective scientific memory, creating immediate value for reproducibility and reuse while laying the foundation for AI-enabled discovery. In this Perspective, we report outcomes of a recent community workshop and propose a practical path toward such an ecosystem. Its benefits unfold along three timescales. In the near term, standardized, searchable, and citable datasets strengthen analysis pipelines independently of AI. In the intermediate future, community-validated tools operating across laboratories could enable broader reuse and increasingly closed-loop automated experiments. In the long term, a diverse corpus of measured electronic spectra could support foundation models for quantum matter. Looking forward, a diverse corpus of measured electronic spectra could support foundation models for quantum matter and new discoveries. We discuss lessons from existing scientific databases, the current state of AI in ARPES, as well as areas of community convergence on data organization, access, credit, governance, and a staged pilot program.
Read the full abstract

All authors Zhi-Xun Shen, Jin Gene Wong, Karen Lei, Leon Garcia, Felix Baumberger, Sergey Borisenko, Aaron Bostwick, Ting Cao, Tanny Chavez, Yulin Chen, Monika Choudhary, Andrea Damascelli, Dan Dessau, Thomas P. Devereaux, Sandy A. Ekahana, Alexei Fedorov, Atsushi Fujimori, Yucheng Guo, M. Zahid Hasan, Makoto Hashimoto, Yu He, Philip Hofmann, Zahid Hussain, Harold Y. Hwang, Zhurun Ji, Chunjing Jia, Timur K. Kim, Phil D. C. King, Takeshi Kondo, Alessandra Lanzara, Chun Lin, Donghui Lu, Matteo Michiardi, Sung-Kwan Mo, Brian Moritz, Laurenz Rettig, Eli Rotenberg, Takafumi Sato, Kyle M. Shen, Jonathan A. Sobota, Inna Vishik, Shuolong Yang, Ming Yi, Timothy Zaklama

Artificial Intelligence

MoMaTune: Model, Match, Then Tune Language Models for Numerical Records

Weimin Lyu, Jingfen Zhang, Lingdao Sha et al. (+4 more)

Despite remarkable progress in language understanding and generation, decoder language models continue to exhibit striking failures on numerical data. Numerical examples are typically serialized as text by concatenating feature names with their values (e.g., fat:3.4), while conventional task adaptation applies loss only to the downstream prediction target. This leaves two structures weakly supervised: within-value sensitivity, where digit position determines numerical magnitude, because token cross-entropy contains no explicit numerical metric; and across-field structure, where measurements in the same record are statistically dependent, because target-only tuning provides no direct loss on the measurements or their dependencies. We introduce MoMaTune (Model, Match, then Tune), a two-stage post-training strategy organized around these supervision gaps. Stage 1 applies Model, ordinary next-token prediction densely over label-free serialized records, and optionally adds Match, a same-record consistency objective that aligns representations of different partial views of the same example. Stage 2 applies Tune, supervised fine-tuning for downstream target prediction. We evaluate MoMaTune on 169,285 records from eight public datasets, with 339 of 340 input fields numerical. At 1.5B parameters, Model raises the eight-dataset masked-value imputation aggregate from 0.106 to 0.401, while the three-dataset correct-context gain rises from 0.058 to 0.167. On diagnostics, Model also increases sensitivity to numerically consequential digit changes, while Match improves robustness to informative partial views. Adding Tune yields higher target-supervised embedding aggregates for each tested Stage-1 parent, and the two-stage recipes obtain higher aggregate scores than Tune alone. The same pattern is observed at 4B, 9.6B, and 27B parameters when model embeddings are used for downstream prediction.
Read the full abstract

All authors Weimin Lyu, Jingfen Zhang, Lingdao Sha, Peng sun, Xiaowei Chen, Vincent Gao, Rui Song

Saturday 10 October

Formal Languages and Automata

A Factor-Two Lower Bound for Complementing Two-Way Deterministic Finite Automata

Sudais Arif

For every $n\ge22$, we construct a language over a fixed three-letter alphabet whose two-way deterministic state complexity is $n$ and whose complement has state complexity $2n$. Acceptance is at the left endmarker, and infinite computations reject. The lower bound holds for arbitrary deterministic recognizers. The proof uses the group of transformations induced by the witness to constrain the behavior of a smaller recognizer, then derives a contradiction from a permutation query problem. A matching halting complement follows from a $2n$-state construction for sources whose destination states determine the direction of each transition. Splitting states by incoming direction gives a $2n+2b$ upper bound for general sources, where $b$ counts states entered in both directions. In the acceptance-anywhere model of Geffert, Mereghetti, and Pighizzini, the same family has exact complexities $N$ and $2N-1$ for every $N\ge23$.
Read the full abstract

Artificial Intelligence

Using Evidence-Centered Design to Guide Prompt Engineering in Measurement: An Illustration with Automated Essay Scoring

Jiawei Xiong, Cheng Tang, Hong Jiao et al. (+1 more)

Generative artificial intelligence (GenAI) models, particularly large language models (LLMs), are increasingly used in educational settings for tasks such as item generation, automated scoring, and feedback delivery. The quality of these outputs depends not only on the AI model but also on how users construct the prompts that guide model behaviors. Current prompt engineering practices are largely ad hoc, relying on trial and error rather than principled design. This lack of structure introduces validity threats including construct misrepresentation, construct-irrelevant variance, and limited reproducibility. We propose that Evidence-Centered Design (ECD), a framework in educational measurement for structuring valid inferences from observed performance, can serve as a principled foundation for prompt engineering in educational applications. We map the three core ECD models, the student model, the evidence model, and the task model, onto the components of educational prompt design, and describe a workflow for ECD-based prompt engineering. We derive five design principles from this mapping and present an empirical study contrasting ECD-designed prompts with state-of-the-art (SOTA) prompting baselines on an automated scoring task. Our empirical study demonstrates the usefulness of this framework on automated scoring. Results suggest that treating prompting as assessment design rather than informal instructions strengthens the validity, reliability, and fairness of AI-generated educational outputs.
Read the full abstract

All authors Jiawei Xiong, Cheng Tang, Hong Jiao, Hanna Choi

Artificial Intelligence

Words to World: Benchmarking LLMs in Interactive Physical Environments

Xinnan Dai, Kaiqi Yang, Yaxin Li et al. (+9 more)

Words are increasingly becoming an interface for action: large language models (LLMs) can now invoke tools, operate devices, and make decisions that affect the physical world. However, current benchmarks largely evaluate LLMs through static text responses, offline tool calls, or simulated tasks, leaving open the question of how well they can act in real physical environments. We present a benchmark for measuring LLMs in closed-loop physical interaction, where models control a harness system connected to real-world components including robotic arms, cameras, and sensors. In each task, the model must issue instructions, receive physical feedback from the environment, interpret the resulting state, and adapt its next action accordingly. This setting evaluates capabilities beyond language understanding, including grounded perception, action planning, tool selection, spatial reasoning, feedback-driven correction, and robustness to execution uncertainty. Through experiments with representative LLMs, we characterize their ability to translate linguistic reasoning into physical actions and identify common failure modes such as hallucinated state changes, poor recovery from unexpected outcomes, and unstable long-horizon plans. Our benchmark aims to provide a systematic foundation for evaluating LLMs as interactive agents that connect words, tools, and the physical world.
Read the full abstract

All authors Xinnan Dai, Kaiqi Yang, Yaxin Li, Cheng Luo, Wanli Li, Mingqing Xiao, Xinyang Jiang, Daoan Zhang, Kaitao Song, Fengze Liu, Zefan Cai, Caihua Shan

Artificial Intelligence

Understanding Unified VLMs: From Attention Probing to Guidance Tuning for Multimodal Generation

Zhengyang Luo, Xiaoyu Qiu, Zhiheng Tan et al. (+3 more)

Unified vision-language models (VLMs) perform both text-to-image (T2I) and image-to-text (I2T) generation with one shared Transformer backbone, yet how the backbone processes the condition in either direction remains poorly understood. We study this question from the perspective of conditional guidance: we probe every attention head to show the activation difference between two branches of classifier-free guidance (CFG) with different conditional guidance. On two unified VLMs of autoregressive and masked-diffusion paradigms, the probing reveals a subset of heads that are more sensitive to the change of condition while others are agnostic to it. This characteristic pattern holds in both T2I and I2T generation and is largely stable across data samples, demonstrating the internal mechanism of unified VLMs. Motivated by this, we propose a new inference-time intervention approach for unified multimodal generation, named HeadTuned-CFG, which uses the probed condition-change awareness to select and tune the conditional guidance at the attention-head level. Experiments show consistent benefits in both multimodal generation directions of unified VLMs without additional training, and in particular substantial improvements over strong baselines in T2I generation. These results further support the key finding of our probing analysis: the observed pattern of functional specialization is an intrinsic property of unified VLMs.
Read the full abstract

All authors Zhengyang Luo, Xiaoyu Qiu, Zhiheng Tan, Yuchen Liang, Yingbin Liang, Liyue Shen

Machine Learning

Zeroth-Order Fine-Tuning of LLMs in Learnable Low-Rank Subspaces

Chuang Tong and Ling Chen

Zeroth-order (ZO) optimization enables memory-efficient fine-tuning of large language models (LLMs). However, ZO optimization suffers from high variance in high-dimensional parameter spaces. While recent methods restrict perturbations to random low-rank subspaces, these subspaces often misalign with task-specific descent directions. We propose AdaSub-ZO, a novel forward-only fine-tuning framework in learnable low-rank subspaces. Instead of relying on fixed random matrices, AdaSub-ZO iteratively learns layer-wise projection matrices directly from finite-difference feedback to track task-aware descent directions. Furthermore, AdaSub-ZO incorporates a subspace momentum acceleration mechanism to smooth noisy gradient estimates across evolving subspace bases. Extensive experiments on natural language understanding tasks demonstrate that AdaSub-ZO outperforms existing ZO baselines, substantially reducing gradient estimation variance and accelerating convergence.
Read the full abstract

Multiagent Systems

Persistent LLM Orchestration of Heterogeneous Robot Teams via Partial Delegation to Multi-Agent Protocols

Maria G. Mendoza, Victoria M. Tuck, Zachary Ravichandran et al. (+2 more)

Natural language specifications let non-expert users task robot teams, and large language models (LLMs) have recently proven effective at grounding such semantic specifications into plans for heterogeneous teams. Existing approaches assume missions that terminate and environments that, while unknown, do not change, but both assumptions fail in many emerging applications. For example, a disaster-response mission does not terminate, and its environment changes continuously as hazards spread and evacuees move. We identify three ways an LLM orchestrator breaks in this setting: its prompt grows without bound, its reasoning degrades as agents and targets accumulate, and it stops actively monitoring after covering the search space once. We build on the successful framework of grounded generative intelligence, and address each issue with a classical idea. A Markovian state summary, rebuilt from the semantic map at every invocation, replaces the growing interaction history. Partial delegation lets the orchestrator assign a subset of robots to an autonomous multi-agent protocol and reserve its reasoning for generating subtasks that require detailed parameters and careful ordering. Lastly, observation staleness, surfaced explicitly in the prompt, together with task-and event-triggered re-prompting, keeps the team vigilant. In a simulated urban search-and-rescue mission with spreading fire and moving evacuees, our framework rescues 98% of civilians versus 89% for the strongest baseline while using six times fewer tokens. Additional ablations isolate the contribution of each mechanism.
Read the full abstract

All authors Maria G. Mendoza, Victoria M. Tuck, Zachary Ravichandran, Shankar Sastry, George J. Pappas

Machine Learning · Cross-listed from math.OC

Non-Euclidean Gradient Descent Has Many Edges of Stability

Peter Zaika and Zakhar Shumaylov

Recent observations that deep learning models train at the edge of stability have renewed interest in understanding stability properties of classical optimization algorithms. For quadratic objectives, standard gradient descent has a sharp stability threshold: it converges from every initialization when $\eta<2/L$, diverges from almost every initialization when $\eta>2/L$, and diverges from every nonzero initialization when $\eta>2/\mu$. For non-Euclidean gradient descent, the corresponding known guarantees are weaker: convergence holds for every initialization when $\eta<2/S$, where $S$ is the generalized sharpness, while every nonzero initialization diverges when $\eta>2/\mu$. The behavior in the intermediate regime $2/S<\eta<2/\mu$ is not fully characterized. We show that Hilbertian norms are the only ones for which $2/S$ is an almost-everywhere sharp threshold for every positive-definite quadratic objective. Using linear conjugacy between non-Euclidean and Euclidean gradient descent, we show that for general norms the intermediate regime can contain arbitrarily many edges of stability associated with distinct attracting directions. We identify a forward-invariant region of exponentially divergent directions beyond the generalized stability threshold. When the directions of minimum curvature are finite and nondegenerate, we prove that, sufficiently close to $2/\mu$ from below, only their invariant directions remain convergent. We give smooth two-dimensional constructions with arbitrarily many distinct thresholds, arbitrarily large relative threshold separation and show these properties to be stable.
Read the full abstract

Data Structures and Algorithms

A Fast Greedy Algorithm with a 0.45 Approximation Guarantee for the Airplane Refueling Problem

Oleg Lomachenko and Arnold Enikeev

We study a natural greedy algorithm for the Airplane Refueling Problem (ARP), based on a well-known priority rule. We show that this algorithm achieves an approximation guarantee of $9/20$, noticeably improving the previously known $1/12$ guarantee for a fast algorithm. We also show that the same priority rule can be implemented in $O(n\log n)$ time and $O(n)$ space using a dynamic convex-hull data structure. These results provide a worst-case approximation guarantee for an existing greedy priority rule and establish an efficient implementation. In the real-RAM model, the running-time bound is independent of the magnitude of the processing times. The resulting method is also suitable as a fast primal routine inside more complex exact and approximation algorithms for large instances. These results narrow the gap between theoretically strong but complex approximation schemes and lightweight algorithms that can be used directly in computational pipelines. The algorithm combines a simple selection rule with a provable performance bound, making it a useful baseline for further algorithmic studies of ARP.
Read the full abstract

Artificial Intelligence

HarnessNet: Feedback-driven Hierarchical Joint Harness Evolution for Multi-agent Systems

Zhaoyuan Qiu, Sachith Seneviratne, Yu Xia et al. (+1 more)

AI agents can evolve by using environment feedback to revise their harnesses, the software managing model inputs, tool use, memory, and execution, while keeping foundation models frozen. Existing studies focus on single agents, leaving joint harness evolution in multi-agent systems (MAS) underexplored. A single global optimizer faces growing context and workload as the system scales, while access restrictions may prevent sharing agent internals or system-wide feedback. We propose HarnessNet, a framework for hierarchical harness evolution in MAS. Its basic unit, Agentron, comprises an evolvable forward harness, a foundation model, and a local optimizer that can observe and modify only that harness. Agentrons form a layered graph through which environment feedback propagates backward, with upstream optimizers receiving node-specific feedback from the downstream layer. This decomposes joint harness evolution into smaller, coordinated tasks with separate contexts, reducing individual optimizers' context and workload while keeping access local. Optimizers need not access all agent trajectories or raw environment feedback. Experiments on AgentBench with multiple foundation models demonstrate the feasibility of training HarnessNet. Compared with a single global optimizer, HarnessNet achieves comparable performance gains while respecting local privacy and requiring fewer tokens and less time per optimizer call. A migration study shows that the evolved harnesses remain jointly effective after replacing the foundation models, suggesting transferability across models.
Read the full abstract

All authors Zhaoyuan Qiu, Sachith Seneviratne, Yu Xia, Saman Halgamuge