Peer Revenue Fully funded

Papers

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

Computer Science

cs Watch Computer Science

Thursday 27 August

Discrete Mathematics

Faster FPRAS for the Permanent via Restricted Poincaré Inequalities and Coupled Flows

Xiaoyu Chen, Eric Vigoda and Xiongxin Yang

The permanent of an $n\times n$ $0/1$ matrix $A$ equals the number of perfect matchings in the bipartite graph with edges defined by $A$. Jerrum, Sinclair, and Vigoda (2004) presented an FPRAS for approximating the permanent of any nonnegative matrix using a novel simulated-annealing algorithm. The running time was improved by Bezáková, Štefankovič, Vazirani, and Vigoda (2008) to $O(n^7\log^4 n)$ for $0/1$ matrices, for any fixed approximation and success parameters. We present the first asymptotic improvement over this running time bound, obtaining an $O(n^6\log^5 n)$-time algorithm. As in the previous works, our algorithm extends to arbitrary nonnegative matrices. The analysis of Bezáková et al. yields an $O(n^4)$ relaxation time bound for the JSV Markov chain on perfect and near-perfect matchings with ideal hole weights, under which each hole pattern (the unmatched vertices, if any) is equally likely in the stationary distribution. We introduce a restricted Poincaré inequality for the partition into hole patterns and prove an $O(n^3)$ bound on the corresponding restricted relaxation time. Our proof uses a coupled multicommodity flow argument inspired by a recent transport-flow argument of Chen et al.~(2025) for the Jerrum-Sinclair chain on all matchings.
Read the full abstract

Computation and Language

Not Just Reason, Not Just Scan: Reinforcement Learning for Proactive Scientific Error Verification over Academic Paper

Rongjin Li, Yuanxin Liu, Hao Zhou et al. (+3 more)

Multimodal large language models (MLLMs) are increasingly capable scientific assistants, yet they remain far from fully autonomous research. This transition requires models to actively inspect academic papers, build global evidence views, and make traceable judgments without prespecified issues or evidence. However, existing work provides limited task paradigms or training studies for such issue- and evidence-absent verification. We study this challenge through scientific error detection, where models must determine whether errors exist and justify them with evidence-based reasoning. To fill this gap, we present VERA-RL, a reinforcement-learning formulation for scientific error detection over academic papers. Following a Reason–Verify–Scan progression, we construct VERA-13K, a 12,900-sample dataset organized into 4,300 matched chains, covering 6 scientific-error categories across the research workflow and broad natural-science domains. We further introduce fine-grained rewards for reasoning completeness, evidence alignment, and error precision. Training Qwen3-VL-8B with VERA-RL substantially improves verifiable reasoning, approaching flagship MLLMs such as Gemini 3 Pro and Qwen3-VL-235B-A22B on Scan.
Read the full abstract

All authors Rongjin Li, Yuanxin Liu, Hao Zhou, Fandong Meng, Jie Zhou, Xu Sun

Systems and Control

Current-Limiting Control for Fault Ride-Through of LLC-based Solid-State Transformer in Data Centers

Haoyu Wang, Chi Zhang, Mafu Zhang et al. (+2 more)

Solid-State Transformers (SSTs) are increasingly proposed as the interface between distribution grids and data centers due to flexible power flows and fast dynamic response. However, when a short-circuit fault occurs in a load branch, the SST with a voltage-source-type DC-DC stage is forced to shut down due to fault currents. Therefore, current-limiting strategies are strongly needed to prevent catastrophic equipment damage and cascading blackouts by instantly restricting massive current spikes and offering sufficient currents for protection devices to act at the faulted branch. This paper proposes a coordinated DC load fault-tolerant current-limiting and recovery strategy embedded directly in the control of the SST DC-DC stage, avoiding additional hardware cost. Specifically, the fault mechanism of an example LLC resonant converter is studied. Accordingly, a fault detection framework is implemented, a closed-loop current controller is proposed to limit the DC current to a designated value within microseconds by surging the switching frequency and adjusting the duty cycle, and a ramped recovery stage will then restore the DC bus after the fault isolation without inrush currents. Experiments on an LLC converter prototype have verified the feasibility of the proposed current-limiting strategy, enabling faster and lower-cost fault response suitable for resilient data center power architectures.
Read the full abstract

All authors Haoyu Wang, Chi Zhang, Mafu Zhang, Rudy Wang, Peter Barbosa

Machine Learning

SimCast-S2S: A Computationally Efficient Diffusion Model for Subseasonal Precipitation Forecasting

Hiep V. Dang and Antonios Mamalakis

Subseasonal-to-seasonal (S2S) precipitation forecasting has substantial financial and societal impact, yet remains challenging because of weak predictive signals, high associated uncertainty, and the computational cost of operational systems, which constrains simulation fidelity. We introduce SimCast-S2S, a generative latent-diffusion framework for probabilistic S2S precipitation forecasting that addresses three major bottlenecks in data-driven prediction. First, because S2S prediction requires uncertainty quantification rather than only deterministic point forecasts, SimCast-S2S is the first data-driven system that uses a diffusion-based generative pipeline for S2S prediction, enabling effective sampling from the underlying conditional distribution. Second, since generating large probabilistic ensembles is computationally costly in physical space, SimCast-S2S instead operates in a compact latent space learned by variational autoencoders (VAEs), enabling efficient large-ensemble generation. Third, diffusion models typically require large training datasets; SimCast-S2S overcomes this via transfer learning with low-rank adaptation (LoRA), pretraining on large ensembles of climate simulations before fine-tuning on limited reanalysis data. On reanalysis data, SimCast-S2S outperforms deep learning baselines, including convolutional neural networks and U-Net architectures. Notably, despite using only a subset of atmospheric input variables and no post-processing, bias correction, or calibration, SimCast-S2S remains competitive with, and in many aspects outperforms, state-of-the-art operational systems such as the ECMWF-S2S baseline. These results indicate that latent generative modeling combined with simulation-to-reanalysis transfer learning offers an efficient and scalable path toward data-driven probabilistic S2S precipitation forecasting.
Read the full abstract

Discrete Mathematics

On identifying codes on oriented graphs

Soura Sena Das and Sagnik Sen

This article studies identifying codes in oriented graphs from a computational complexity perspective. We investigate the $\mathcal{F}$-Id Code problem, where given a simple graph $G$ and a vertex subset $C$, which induces a subgraph in the family $\mathcal{F}$, as inputs and ask whether it is possible to orient $G$ in such a way that $C$ becomes its oriented identifying code. Focusing on the family $\mathcal{F}_d$ of $d$-regular graphs, we establish a complete dichotomy by proving that the problem is polynomial-time solvable for $d\leq1$ and NP-complete for all $d\geq2$.
Read the full abstract

Computation and Language

Benchmarking Clinical Decision Pathway Adherence in Large Language Models

Nuo Chen, Xinyang Jiang, Zilong Wang et al. (+5 more)

Following clinical decision pathways (CDPs) defined by clinical practice guidelines is essential for safe and reliable medical decision-making. However, existing medical large language model (LLM) benchmarks mainly evaluate final-answer accuracy, providing limited evaluation of models' ability to adhere to guidelines. To address this gap, we introduce MEGA-CDP, a benchmark for evaluating whether medical LLMs can generate guideline-adherent CDPs using provided guidelines as references. MEGA-CDP is constructed from 2,274 English and Chinese clinical practice guidelines through a guideline-to-case pipeline, yielding 42,353 clinical cases with explicit reference CDPs. It supports both single-turn vignette and multi-turn interactive settings, and introduces a CDP-oriented evaluation framework for measuring pathway consistency. Experiments on 16 representative LLMs show that reliable clinical decision support remains challenging for current models, demonstrating the need for CDP-oriented evaluation and the value of MEGA-CDP for advancing guideline adherence in medical LLMs.
Read the full abstract

All authors Nuo Chen, Xinyang Jiang, Zilong Wang, Zhifei Zhang, Xiaoye Qu, Jiajun Deng, Yulan Guo, Cairong Zhao

Computer Vision

DPA-I2P: Depth-Guided Projective Alignment for Image-to-Point-Cloud Registration in Autonomous Driving

Wenxin Zhang, Hang Li, Zhiwei Xu et al. (+3 more)

Image-to-Point Cloud Registration aims to estimate the camera pose of a given image within a 3D scene point cloud, which is a fundamental task in autonomous driving and large-scale outdoor localization. Recent implicit correspondence learning methods have improved registration performance by learning cross-modal alignment in an end-to-end framework, leading to more accurate camera pose estimation. However, due to the inherent modality discrepancy between images and sparse LiDAR point clouds, reliable cross-modal correspondence learning remains challenging. To address this issue, we propose Depth-Guided Projective Alignment for Image-to-Point-Cloud Registration (DPA-I2P). Unlike naive depth or feature concatenation, Ray-Conditioned Metric Depth Encoding (RMDE) and Projection-Consistent Vision Lifting (PVL) exploit depth and visual cues in a structured, geometry-aware manner. In addition, Cross-Modal Query Pruning (CQP) suppresses unreliable queries during early refinement to improve matching stability. Experiments on KITTI and nuScenes demonstrate the effectiveness of the proposed method. On KITTI, DPA-I2P reduces RTE and RRE by 45.0% and 55.6% over the strongest implicit baseline, respectively. On nuScenes, DPA-I2P also improves registration accuracy over the evaluated baselines, suggesting better transferability to different driving scenes.
Read the full abstract

All authors Wenxin Zhang, Hang Li, Zhiwei Xu, Qiankun Dong, Gang Wang, Tao Li

Cryptography and Security

Unsaid, Unsafe? Implicit Security Obligations in LLM-Based RTL Code Generation

Guang Yang, Xing Hu, Xiang Chen et al. (+1 more)

Large Language Models (LLMs) generate register-transfer-level (RTL) code with rapidly improving functional correctness. Security of LLM-generated code, however, has been studied mainly for software, where flaws can still be patched after deployment. Insecure RTL offers no such remedy once taped out into silicon. We construct SECRTL-GEN, a multi-language resource-access security benchmark grounded in real SoC IP: 392 tasks over five CWE families and four HDLs (Verilog, SystemVerilog, VHDL, and Python), each with black-box functional and security testbenches. Functional specifications intentionally omit security obligations, matching how obligations are often kept out of functional docs in practice. An empirical study of five frontier LLMs shows a sharp gap: under vanilla prompts they pass functional tests in about 73-79% of cases but security tests in only 14-35%, and stronger functional models are not safer. Adding CWE knowledge raises security, while unaided self-thinking helps less and both security-oriented prompts cut functional pass rates, showing that the bottleneck is missing weakness awareness in the specification, not an inability to write defensive RTL. We present RTL-Obliger, a neuro-symbolic framework that infers these implicit obligations. An LLM extracts a functional-semantic graph from the specification; a symbolic engine then matches it against a CWE pattern ontology to surface mitigation-evidence gaps and signal-level obligations; the LLM finally revises RTL under those obligations in a functionality-preserving two-stage generation. Across five models and four languages, RTL-Obliger raises mean all-pass from 49.6-51.4% (SecV/RESCUE) to 61.6%, with higher security and functional rates than these secure-generation baselines.
Read the full abstract

All authors Guang Yang, Xing Hu, Xiang Chen, Xin Xia

Computation and Language

Surgical Alignment in Knowledge Graph Training for Clinical Diagnosis with Large Language Models

Saksham Khatwani, He Cheng, Majid Afshar et al. (+2 more)

Biomedical knowledge graphs (KGs) offer structured medical knowledge that can ground large language model (LLM) reasoning in clinical diagnosis application, yet how KG signal should be integrated into LLMs remains an open question. We present a systematic study spanning five KG task formulations, three training paradigms, two KGs, and three base LLMs. At the task level, all paradigms improve over the non-finetuned baseline, but methods with comparable in-domain accuracy show substantially different knowledge transfer behavior. We introduce Gradient Intervention Density (GID) and Gradient Distortion (GD) to measure how broadly an optimizer modifies the pretrained model. GID and GD together reveal a clear divide: KG-judgment training under KL regularization produces sparse, localized updates (a regime we term as surgical alignment), while task-specific SFT produces dense ones. A controlled ablation shows that the objective and KL contribute to sparsity independently, and the paradigms that produce sparse updates also improve reasoning quality, even when their in-domain accuracy is lower than task-specific SFT. Assessing KG-LLM integration thus requires complementing accuracy with optimization-geometry diagnostics. Our implementation can be found at https://github.com/LARK-NLP-Lab/Surgical-Alignment.
Read the full abstract

All authors Saksham Khatwani, He Cheng, Majid Afshar, Dmitriy Dligach, Yanjun Gao

Machine Learning

VGAS: Variance-Reduced Guidance and Adaptive Selection for Training-Free Reward Alignment in Discrete Diffusion

Kwanyoung Kim

Masked discrete diffusion models perform strongly on text, code, and biological sequences, but their training objective rewards only naturalness, and retraining the generator for every new reward is expensive. Inference-time steering of a frozen model either guides the sampler by the reward gradient or searches over several trajectories, and recent samplers combine the two. Such combinations are assembled as pipelines that leave three choices at their defaults: a guidance estimate resting on one Gumbel draw per sample, a reward tilting placed without reference to the distribution the combination then targets, and a selection temperature held fixed although the spread of per-step rewards drifts. We identify that distribution and settle the three choices against it. We therefore propose Variance-reduced Guidance and Adaptive Selection (VGAS), a simple yet effective inference-time framework that reduces the variance of the guidance estimate for both reward types, applies the reward tilting in the clean-token logits, where the pretrained schedule is preserved, and sets the selection temperature per step. Across regulatory DNA, protein and small-molecule benchmarks, VGAS attains the best training-free reward and matches or surpasses a reward-fine-tuned generator.
Read the full abstract

Computers and Society · Cross-listed from econ.EM

DIRECT: Decomposing Audience Preference and Creative Effect in Visual Content Analytics

Yizhi Liu, Balaji Padmanabhan and Siva Viswanathan

Which visual choices make a post perform better? A growing literature answers this question with pooled coefficients estimated across many creators, which platforms translate into creative recommendations. We show that these coefficients blend two distinct patterns that can point in opposite directions for the same attribute. The first, audience preference, arises because creators who favor a style attract differently composed audiences, so their posts perform differently because of who is watching, not what any single post does. The second, creative effect, captures how a creator's audience responds when she departs from her usual look. Pooled estimation averages the two, and audience preference can be large enough to reverse the signal that creative direction requires. We propose DIRECT (Decomposed Identification of Response Effects via Causal Tools), a panel-based causal-inference framework that separates them, combining the Mundlak between-within decomposition with double machine learning over latent vision-language treatments that co-vary within a creator. We apply it to 232,088 sponsored Instagram beauty posts across 1,527 creators and 11 CLIP-derived visual style axes. The two carry opposite signs on 4 of 11 attributes, and on 2 of 11 the pooled coefficient itself recommends the wrong creative direction: on skin tone, it favors lighter representations while the creative effect points the other way, since a creator's audience engages more with tones darker than her baseline. On held-out creators, prescribing from the pooled coefficient forgoes 31% of the achievable engagement gain. We contribute a diagnosis of estimand mismatch in visual content analytics, a framework that recovers the decision-relevant estimand from observational panel data, and three portable diagnostics for auditing whether pooled estimates support the decisions they inform.
Read the full abstract

Robotics

SOLO: Stable Omni-terrain Long-Horizon Perceptive Humanoid Locomotion

Pihai Sun, Gang Han, Jingkai Sun et al. (+17 more)

Humans traverse complex terrain over long distances without losing balance, whereas perceptive humanoid policies become fragile as perception and control errors accumulate. We present SOLO, a unified framework addressing two compounding causes of this long-horizon fragility: dense terrain reconstruction smooths action-critical details, and pointwise imitation lacks temporal credit assignment. Its Query Reconstructor (QR) uses Fourier-encoded cell queries to retrieve spatially specific evidence from depth-proprioception tokens, preserving sharp terrain boundaries. Trajectory-Aware MSE (TA-MSE) Distillation adds next-state teacher-student disagreement to the PPO reward, enabling Generalized Advantage Estimation to propagate future disagreement penalties to preceding actions. In simulation, QR reduces height-map L1 error by factors of 3.3-4.0, while TA-MSE surpasses PPO and MSE+PPO in curriculum progression. On stress-test terrains, SOLO achieves 97.5% mean traversal success and 96% stepping-stone success, versus 75.0-75.6% and 0-3% for dense-reconstructor variants. Deployed zero-shot with only a chest-mounted depth camera and proprioception, SOLO completes a continuous 1.5-km outdoor route and an indoor mixed-terrain course. Project page: https://sunpihai-up.github.io/solo/
Read the full abstract

All authors Pihai Sun, Gang Han, Jingkai Sun, Jiahao Ma, Zeran Su, Zelin Tao, Peiran Liu, Shuai Shi, Wei Cui, Zifan Wang, Jialin Yu, Wen Zhao, Kangning Yin, Jiaxu Wang, Jiahang Cao, Lingfeng Zhang, Hao Cheng, Jian Tang, Qiang Zhang, Yijie Guo

Machine Learning

J-Zero: Unified Challenger–Solver–Judge Self-Evolution from Zero Data

Gyouk Chu, Myeongho Jeon, Teresa Yeo et al. (+1 more)

Self-evolving language models have recently emerged as a promising path toward superintelligence, with the advantage of reducing the cost of human supervision. While considerable progress has been made in verifiable domains, self-evolution in unverifiable domains remains less explored. We propose Judge co-adaptation from Zero data (J-Zero), a unified Challenger–Solver–Judge self-evolution framework that supports self-improvement across both domains. The Challenger and Solver co-evolve through an adversarial interaction: the Challenger generates increasingly difficult tasks, while the Solver learns to produce higher-quality responses to them. In parallel, the Judge co-adapts using preference pairs whose ordering is known in advance from how each response was produced, i.e., the Solver's answer over the Challenger's, and the Solver's decomposed-and-recombined answer over its one-shot answer, rather than from the Judge's own scores. J-Zero outperforms the baselines by an average of 4.2 points on verifiable and 8.0 points on unverifiable domains, and continues to improve through at least ten iterations, whereas the baselines degrade after two. Further analysis identifies Judge co-adaptation as the key driver of this sustained improvement.
Read the full abstract

All authors Gyouk Chu, Myeongho Jeon, Teresa Yeo, Eunho Yang

Machine Learning

Activation Outliers Matter: Robust Recovery for Quantized Multimodal LLMs

Tanzila Rahman, Mehran Taghian Jazi, Yunke Peng et al. (+10 more)

Low-bit quantization offers a promising avenue for reducing the computational and memory demands of Multimodal Large Language Models (MLLMs). Recent hardware support for low-precision formats, ranging from MXFP8 to ultra-low-bit formats such as MXFP4 and HiF4, has accelerated research into efficient MLLM training and deployment. In this work, we present a systematic study of these quantization schemes in representative MLLMs that span both video generation and reasoning tasks. Our analysis shows that MXFP8 achieves near-lossless performance, whereas aggressive 4-bit quantization leads to significant degradation. Through extensive ablations, we identify activation quantization as the primary source of this performance loss, contributing substantially more than weight quantization. Motivated by this observation, we propose Residual Fallback Quantization (RFQ), a lightweight activation reconstruction framework that supplements the primary ulta-low-bit activation representation with an auxiliary quantized residual pathway. By explicitly modeling and compensating for quantization errors, RFQ improves activation fidelity while preserving the efficiency advantages of ultra-low-bit computation. RFQ requires no architectural modifications and incurs negligible computational overhead. Extensive experiments on Wan2.2 and Qwen3-VL demonstrate that RFQ consistently recovers a substantial portion of the performance lost under the quantization of MXFP4 and HiF4, significantly narrowing the gap to BF16 baselines across both generation and 4 reasoning benchmarks. Our findings establish activation quantization as the dominant bottleneck in ultra-low-bit MLLMs and highlight residual-based activation reconstruction as an effective and practical strategy for robust 4-bit deployment.
Read the full abstract

All authors Tanzila Rahman, Mehran Taghian Jazi, Yunke Peng, Zhuang Ma, Anandharaju Durai Raju, Yao Wang, Xing Huang, Hei Yi Mak, Shadan Golestan, Hoang Le, Yonghan Dong, Wei Guo, Yaoyuan Wang

Computer Vision

Visual Information-Guided Parallel Decoding for Diffusion Multimodal Large Language Models

Insu Lee, Wooje Park, Wonseok Shin et al. (+2 more)

Diffusion multimodal large language models (dMLLMs) have recently emerged as a new decoding paradigm for multimodal generation. Starting from a fully masked sequence, dMLLMs progressively decode the sequence by unmasking a subset of the remaining masked positions at each step. Since the selected tokens serve as the prediction context for subsequent steps, deciding which tokens to decode is crucial to the quality of the final output. The most common strategy prioritizes tokens based on a certainty measure that tends to favor tokens frequently observed in the training data. Recent approaches instead order tokens according to their influence on subsequent predictions, but do not explicitly account for the input image. We propose the Visual Information-Guided Sampler (VIG-Sampler), which prioritizes tokens based on their attention to image tokens. We further impose a constraint that penalizes candidate tokens whose image-attention distributions are similar to those of previously selected tokens, thereby increasing the information gain of the decoded subset. Extensive experiments on 7 captioning and VQA benchmarks with 3 open-source dMLLMs demonstrate the effectiveness of VIG-Sampler, which outperforms the Info-Gain Sampler by an average of 19.3 CIDEr points across the captioning benchmarks and surpasses it on COCO Caption while using only half as many decoding steps.
Read the full abstract

All authors Insu Lee, Wooje Park, Wonseok Shin, Jinwoo Son, Byonghyo Shim

Information Retrieval

Preference Flow Matching with Spectral Factorization for Micro-video Recommendation

Xinxin Dong, Haokai Ma, Fei Hu et al. (+4 more)

Micro-video recommendation aims to infer user preferences from historical interactions and multimodal video content, thereby identifying the next video of interest. However, prevailing methods compress frame sequences into a single holistic representation, entangling the stable visual semantics and the evolving dynamics that jointly shape user preferences. Meanwhile, diffusion- and flow matching-based recommenders condition their generation process solely on coarse behavioral context, leaving its internal temporal structure outside preference formation. We therefore propose PrismRec, a Preference Flow Matching framework with Spectral Factorization for Micro-video Recommendation. Analogous to a prism that disperses white light into its constituent spectrum, PrismRec devises Spectral Semantic Factorization (SSF) to derive complementary static semantic and dynamic factors from frame-level representations via a prior-guided learnable frequency mask in the temporal frequency domain. Then, it proposes Context-Calibrated Preference Matching (CPM) to weigh them with each user's specific sensitivity and inject the calibrated context as a structured condition to steer the matching trajectory toward the target representation, making video content as an intrinsic driver of preference formation rather than auxiliary side information. Experiments on four datasets from two platforms show that PrismRec surpasses the SOTA baseline by up to 22.65%, with the lowest inference cost and peak memory among the compared methods.
Read the full abstract

All authors Xinxin Dong, Haokai Ma, Fei Hu, YuZe Zheng, Bin Wu, Yonghui Yang, Xiaodong Wang

Robotics

TrapVLA: Trapping Vision-Language-Action Models in Configured Failure Modes

Jun-Hui Liu, Kun-Yu Lin, Yi-Lin Wei et al. (+9 more)

This work introduces Configured Failure Trapping, a novel backdoor attack task against Vision-Language-Action (VLA) models, which aims to activate attacks through stealthy textual triggers and induce configured failure modes. Unlike prior backdoor attacks that treat any task failure as a successful attack, Configured Failure Trapping requires the attacker to control how the robot fails (e.g., causing the robot to grasp with a specified positional offset), making it substantially more challenging and hard to detect. To support the new task, we propose an effective data engine for synthesizing high-quality target trajectories and an automated suite for measuring configured-failure fidelity. Then, based on this foundation, we construct two new benchmarks, namely Trap-LIBERO and Trap-RoboTwin, that instantiate Configured Failure Trapping across four representative failure modes. To address this task, we identify sparse action deviation as a critical challenge and accordingly propose a novel method named TrapVLA, which explicitly learns trigger-induced action residuals to steer the policy toward the configured failure behavior. Extensive experiments across simulation benchmarks and real-world robotic settings show that TrapVLA effectively injects configured failure modes into VLA models while largely preserving performance on clean data. Project page: https://john-liua.github.io/TrapVLA/
Read the full abstract

All authors Jun-Hui Liu, Kun-Yu Lin, Yi-Lin Wei, Xu-Han Chen, Yinghao Li, Zhuohao Li, Yuan-Ming Li, Qing Zhang, Xiaoyi Fan, Dongmei Jiang, Yan Li, Wei-Shi Zheng

Computation and Language

Double Trouble: Bilingual Pretraining Leaves Language-Conditioned Effects in Shared-Language Representations

Anjishnu Mukherjee, Ziwei Zhu and Antonios Anastasopoulos

A concept can carry different associations across languages, while modern language models learn English alongside many other languages during pretraining. Yet comparisons among existing models cannot easily isolate how any one language changes the way these models represent English concepts because their training corpora, compute, architectures, and random seeds all differ. We study this question through a controlled experiment with 40 matched 310M-parameter decoder-only models that share an architecture, tokenizer, training recipe, and English data source. Each bilingual condition adds one of eight languages, while four experimental comparisons separately account for English exposure, total training, and English-document overlap. We align each model pair using 3,000 common English words, then measure where 1,000 held-out English concepts fall along 50 fixed semantic contrasts, such as red versus white. Across 32 experimental comparisons, English concept positions differ more between bilingual and English-only conditions than between English-only runs with different random seeds. These differences are larger in contextual states than in token embeddings and peak in middle layers. The language learned alongside English can therefore change how a model represents English concepts even when its English input representations are explicitly aligned.
Read the full abstract

Distributed Computing

Benchmarking Confidential Computing Performance on NVIDIA Blackwell GPUs

Amean Asad and Ansgar Grunseid

This paper measures the performance impact of running large language model inference and training inside a Trusted Execution Environment (TEE) on NVIDIA B200 GPUs, using Intel Trust Domain Extensions (TDX) confidential VMs together with NVIDIA Confidential Computing (CC) on Blackwell GPUs. The performance impact is derived from paired confidential versus non-confidential runs on a single physical host where the only variable is the GPU CC bit and the TDX guest object in the VM launch. The main result is that confidential inference on Blackwell achieves low single-digit throughput overhead when the stack is configured correctly, at about 1-3%. Stock inference stacks incur 30 to 40% penalties due to avoidable configurations rather than the achievable operating point. The cost is not fully represented by a single number because it is governed by two independent axes, a fixed per-host-operation cost that amortizes as batch size grows and a per-NVLink-traffic cost that tracks the share of the step spent in encrypted collectives, and which of the two dominates is set by the workload and the software. We localize each cost to a specific encrypted boundary, give a microbenchmark that predicts the serving penalty to within a submission count, and end with concrete deployment guidance. GPU compute, energy draw, and usable memory capacity are unaffected by CC.
Read the full abstract

Computation and Language

Dependency-Aware Revocable Decoding for Efficient Diffusion Large Language Model Inference

Wooje Park, Insu Lee, Minyoung Noh et al. (+4 more)

Diffusion large language models (dLLMs) offer a promising alternative to autoregressive generation by decoding multiple tokens in parallel through iterative denoising. However, increasing decoding parallelism often degrades generation quality, as early errors can contaminate later contexts. Revocable decoding mitigates this issue by re-evaluating decoded tokens and remasking unreliable ones, but existing methods overlook that unreliable tokens may also corrupt the verification context itself. We identify this failure mode and propose Dependency-Aware Revocable Decoding (DARD), a training-free framework that separates tokens into masked, candidate, and unmasked states. DARD verifies candidate tokens using a selective context that excludes less reliable tokens and adaptively regulates their influence on subsequent decoding. Experiments across 12 textual and multimodal benchmarks on 3 open-source dLLMs show that DARD consistently improves the speed-quality Pareto frontier over recent revocable decoding methods, achieving a 2.71$\times$ speedup and a 4.35-point CIDEr score gain over Saber on Flickr30K.
Read the full abstract

All authors Wooje Park, Insu Lee, Minyoung Noh, Jaeyun Jang, Sungmin Lee, Kyuhong Shim, Byonghyo Shim

Distributed Computing

Optimizing API Gateway Placement in Multi-Cloud Kubernetes

Vinoth Punniyamoorthy, Murali Shankar Dulam, Aswathnarayan Muthukrishnan Kirubakaran et al. (+5 more)

The use of API gateways within geographically distributed multi-cloud Kubernetes clusters poses a tradeoff between infrastructure cost, computational resources, and network latencies. We present an optimization formulation that addresses API gateway placement as a capacitated facility location problem that jointly determines which candidate clusters to activate, how many gateway replicas to deploy, and how regional traffic should be distributed across the selected clusters. The formulation imposes an upper bound on estimated client-to-cluster network round-trip latency, excluding gateway processing, queuing, and backendservice latency, and incorporates a utilization headroom factor for gateway replica capacity. We present both a mixed-integer linear programming (MILP) formulation and a constructive greedy heuristic that ranks candidates according to incremental cost, comprising cluster-activation and marginal replica costs, per unit of assignable capacity while explicitly accounting for already committed load. Both formulations are applied to deterministic, seed-controlled, geography-based synthetic instances. For each problem size, 30 instances are generated with random seeds to analyze their performance. The greedy algorithm achieves an optimality gap of 3.2% to 4.7% to the MILP optimal solution, with a maximum observed gap of 25.0% for one particular instance, and a speedup of approximately 660x to 3,490x for 3 to 12 candidate clusters. In a canonical 10-candidate, 10-demand region instance, MILP-optimal deployment saves 24.2% in terms of monthly cost compared to the full-replication baseline. On the other hand, selecting the single cheapest candidate yields savings of 24.8% compared to the MILP optimum but does not satisfy the latency requirement for 3 out of 10 demand regions.
Read the full abstract

All authors Vinoth Punniyamoorthy, Murali Shankar Dulam, Aswathnarayan Muthukrishnan Kirubakaran, Akshay Deshpande, Nachiappan Chockalingam, Bikesh Kumar, Naga Surya Pasupuleti, Narender Reddy Bitla

Machine Learning

Arrive and Survive: Scaling Safe Goal-Conditioned Policy Learning from One-Bit Failure Signals

Guopeng Li, Yiyang Duan, Yiru Jiao et al. (+1 more)

Contrastive reinforcement learning (CRL) scales effectively in goal-conditioned tasks by casting policy learning into a self-supervised contrastive objective. However, in a failure-terminated Markov decision process, established CRL considers pre-failure future goals only when constructing positive samples, without accounting for the probability mass removed by failure termination. Our theoretical analysis shows that this omission induces a systematic overestimation bias in goal-reaching values. Consequently, near-failure trajectories provide disproportionately strong supervision of success despite retaining little future occupancy. Unsafe actions can thereby be reinforced through catastrophic failure bootstrapping, leading to failed policy learning and unsustainable goal-reaching behaviours. To address this problem, we introduce two minimal yet strong corrections: mass-weighted InfoNCE corrects the overweighting of short surviving futures in critic learning, and a log-survival-mass score restores the missing survival mass in policy optimization. The resulting method, Safe Contrastive Reinforcement Learning (Safe-CRL), requires only the one-bit signal provided by failure termination to scale safe goal-conditioned policy learning. Across twelve failure-prone robot navigation and locomotion tasks, Safe-CRL consistently improves survival and substantially outperforms the Scaling-CRL baseline in goal-reaching performance. Additionally, deep Safe-CRL policies exhibit complex failure-avoidance behaviours. This study completes the CRL theory under failure termination and provides a scalable safe RL framework. The code is available via https://github.com/RomainLITUD/safe-crl.
Read the full abstract

All authors Guopeng Li, Yiyang Duan, Yiru Jiao, Chengcheng Xu

Human-Computer Interaction

EmoSay: Artificial Intelligence-Driven Text-to-Emotional-Speech System for Affective Communication in Extended Reality

Sikiru Ademola Adewale, Sunday D. Ubur, Nikitha Donekal Chandrashekar et al. (+2 more)

While contemporary neural text-to-speech (TTS) systems have achieved high levels of intelligibility, they frequently lack the emotional nuance required for authentic affective communication. This limitation is particularly critical in Extended Reality (XR), where the absence of emotionally expressive audio can diminish user presence and spatial immersion. We present EmoSay, an Artificial Intelligence-driven Text-to-Emotional-Speech (TTES) system designed to bridge the semantic-affective gap in immersive environments. EmoSay modulates a neural synthesis pipeline using discrete emotional prompts, delivering the output through a Unity-based interface featuring high-fidelity spatialized audio. The system was evaluated through a comprehensive user study focusing on perception, engagement, and the subjective sense of empathy. Our results demonstrate that EmoSay significantly enhances the immersive experience, achieving a System Usability Scale (SUS) score of 74.76, indicating strong usability and seamless integration within the XR workflow. Subjective assessments reveal a high degree of perceived naturalness and a strong positive correlation between emotional expressiveness and user engagement. Regression analysis identifies vocal naturalness as the strongest of the tested predictors of user satisfaction, suggesting that EmoSay's affective prosody helps meet the heightened expectations for realism in immersive settings. This work contributes a scalable, affect-aware framework for inclusive XR design and demonstrates the role synthetic emotion can play in fostering human-computer rapport through voice-first interaction.
Read the full abstract

All authors Sikiru Ademola Adewale, Sunday D. Ubur, Nikitha Donekal Chandrashekar, Onyeka Emebo, Denis Gračanin

Computation and Language

SPT: Skills as Pre-Training Data for Agentic Language Models

Yufei Sun, Yudong Li and Yiming Cheng

Agentic (tool-using) language models are mainly trained on tool-call traces and agent trajectories during post-training. These data provide direct behavioral supervision, but producing them requires task environments, execution, and verification, making broad tool and task coverage expensive. Publicly available skills offer another source of training data: they encode reusable tool semantics and workflows but are typically used only as inference-time context. We introduce Skill Pre-Training (SPT), a mid-training method that applies causal language modeling to SkillCorpus, a collection of public multi-file skill packages, optionally mixed with general data. To preserve relations among files within each package, we also introduce Reference Insert, a reference-aware assembly strategy that places supporting files near their mentions in the primary instruction. Experiments across multiple model scales and post-training recipes show that SPT consistently improves agentic performance over mid-training on general or trajectory data, while largely preserving general performance. Data mixture experiments show additional benefits from combining skill data with general annealing corpora. These results indicate that skill packages are a valuable data source for pre-training agentic language models.
Read the full abstract

Networking

SFC-Aware Online Aggregated Data-Link Orchestration for SDN/NFV-Enabled SAGINs

Ziyang Guo and Bing Du

Civil aviation space-air-ground integrated networks (SAGINs) are expected to support heterogeneous cockpit and cabin services over dynamic air-to-air (A2A), air-to-ground (A2G), and air-to-satellite (A2S) data links. This paper studies service function chain (SFC)-aware online access-side aggregated data-link orchestration for civil aviation SAGINs enabled by software-defined networking and network function virtualization (SDN/NFV). We jointly orchestrate spatial bearer resources and temporal elasticity enabled by temporal elastic mapping and parking (TEMP). A rolling-slot model is developed with four request-level actions: NOW, TEMP, REJECT, and DROP, under a lexicographic objective that prioritizes service success, then normalized access-orchestration delay, and finally residual TEMP-related risk. To avoid exhaustive search over the full binary action space, we propose Model-Induced Risk and Scarcity-Aware Refinement (MRSAR), which derives deferred-realization risk from the success-prioritized objective and resource-scarcity signals from data-link and TEMP buffer constraints. These signals guide valuation-based feasible construction and bounded neighborhood refinement. Simulation results show that MRSAR remains close to the Gurobi-MILP reference in service success, outperforms arrival-order and delay-myopic greedy baselines, controls REJECT and DROP failures, and achieves a favorable quality-complexity tradeoff for rolling online orchestration.
Read the full abstract

Software Engineering

DeepRepro: State-Aware Subplanning for Paper-to-Code Reproduction in Evolving Repositories

Hongru Song, Ruqing Zhang, Jiafeng Guo et al. (+2 more)

Recent advances in agentic large language models (LLMs) have enabled increasingly autonomous software engineering workflows, yet automatic machine learning (ML) paper-to-code reproduction remains a challenging long-horizon problem. Unlike conventional code generation, this task requires constructing and maintaining a fully functional repository whose state continuously evolves during execution. Existing systems typically rely on static upfront planning followed by sequential file-level generation, which often leads to inconsistencies as dependencies, interfaces, and execution feedback change over time. We propose DeepRepro, a state-aware framework for paper-to-code reproduction based on execution-state-aware subplanning. DeepRepro dynamically transforms evolving repository states and runtime feedback into fine-grained implementation subplans, keeping planning aligned with execution throughout repository construction. The framework further incorporates repository-aware orchestration and a lightweight process-aware interface for transparent monitoring of long-horizon reproduction. Experiments on PaperBench Code-Dev show that DeepRepro consistently outperforms strong scientific and commercial code-agent baselines.
Read the full abstract

All authors Hongru Song, Ruqing Zhang, Jiafeng Guo, Xueqi Cheng, Maarten de Rijke

Machine Learning · Cross-listed from cond-mat.dis-nn

Dynamical phase selection controls compute scaling in looped transformers

Gunn Kim

A looped transformer performs inference by iterating a weight-tied map, making its computation a dynamical process whose cost is set by the resulting inference dynamics. Here we show that networks with identical architecture and objective, trained to identical accuracy, nevertheless realize distinct dynamical phases depending strongly on initialization, and that the bifurcation defining each phase determines how test-time compute scales. The phases are distinguished by their bifurcation mechanisms, including a saddle-node fold and a Neimark-Sacker-type transition to bounded nonstationary motion. In the fold phase, a one-dimensional normal-form reduction predicts both the relaxation-time and spectral-gap amplitudes from local derivatives of the trained map, yielding the parameter-free relation $\tau(\varepsilon)[1-\lambda_{\max}(-\varepsilon)]\to\pi$. Composed with a regular distribution of problem difficulty, the same critical slowing down produces the workload-level tail $P(\tau>N)\sim N^{-2}$. In the Neimark–Sacker phase, the fold scaling law disappears rather than merely changing its prefactor. Thus, test-time compute is not determined by architecture alone. It is governed by the dynamical phase of the solution found by training.
Read the full abstract

Data Structures and Algorithms

Hadamard Flattening and Gaussian Pooling Sketch for Least Squares with Coordinate-wise Guarantee

Zhao Song and Lichen Zhang

Randomized sketch-and-solve algorithms accelerate overconstrained $\ell_2$ regression by replacing the input with a much smaller, randomly projected problem. While standard subspace embeddings nearly preserve the regression cost, coordinate-wise accuracy is more delicate: rather than the objective value, we want the solution vector itself to be close to the optimum in $\ell_\infty$ norm. Specifically, we seek $x'\in\mathbb{R}^d$ with $\|x'-x^\star\|_\infty\leq\frac{\epsilon}{\sqrt d}\|Ax^\star-b\|_2\|A^\dagger\|_{\mathrm{op}}$, where $A\in\mathbb{R}^{n\times d}$ is the design matrix, $b\in\mathbb{R}^n$ the label vector, $x^\star$ the optimal solution and $A^\dagger$ the pseudo-inverse of $A$. Price, Song and Woodruff introduced this problem and claimed that the subsampled randomized Hadamard transform (SRHT) with $O(\epsilon^{-2}d^{1+\Theta(\sqrt{\log\log n/\log d})})$ rows achieves this guarantee; we point out a gap in the proof. Song, Ye, Yin and Zhang later claimed $O(\epsilon^{-2}d\log^3 n)$ rows, but their proof relies on an independence assumption that does not hold in general, and we exhibit an explicit instance on which it fails. To achieve a truly nearly-linear-in-$d$ row count, we introduce a new fast, dense randomized transform combining a randomized Hadamard flattening, a random permutation, and balanced, disjoint Gaussian pooling. Conditioned on the Hadamard-and-permutation stage, the sketched problem becomes an exact Gaussian regression whose noise is independent of the entire sketched design, exactly what Song, Ye, Yin and Zhang's argument was missing. Our sketch yields the $\ell_\infty$ guarantee with $m=O(\epsilon^{-2}d\log d)$ rows, using one Hadamard pass with padded internal dimension $N=\widetilde{O}(n+\epsilon^{-2}d^3)$. It is also efficient to apply: $(SA,Sb)$ can be computed in $O(Nd\log N)=\widetilde{O}(nd+\epsilon^{-2}d^4)$ time.
Read the full abstract

Systems and Control

Nonlinear Model Predictive Control for Guidance Law with Target Input Estimation

Minho Jang, Minjeong Kim and Sungsu Park

This paper presents a look angle-based nonlinear model predictive control guidance (MPCG) method for missiles equipped with strapdown seekers. Conventional proportional navigation guidance (PNG) requires line-of-sight (LOS) rate measurements, which are not directly available in strapdown systems. MPCG instead employs look angles and their derivatives as state variables, eliminating body-rate coupling and associated parasitic feedback. The guidance problem is formulated as a continuous-time optimal control problem (OCP), discretized via the Legendre-Gauss-Radau pseudo-spectral method (LGRPM), and solved as a nonlinear program (NLP) incorporating explicit field-of-view (FOV) and acceleration constraints. Target acceleration at the first step of the prediction horizon is estimated using an adaptive extended Kalman filter (AEKF) integrated with an interacting multiple model (IMM) framework. Simulation results under single-maneuver scenarios, which include pitch and yaw plane weaving as well as barrel-roll maneuvers, demonstrate that MPCG achieves reliable interception while satisfying operational constraints, outperforming pure PNG (PPNG) in stability and resilience. This indicates that MPCG offers a practical and effective solution for modern missile guidance systems constrained by seeker measurement limitations.
Read the full abstract

Computation and Language

SPEAR: Distilling Domain-Adaptive Reasoning Skeletons via Sequential Symbolic Alignment in Reinforcement Learning

Zhuochun Li, Yuelyu Ji, Yiming Zeng et al. (+1 more)

Reinforcement learning-based knowledge distillation has the potential to transfer complex reasoning from teacher to student models, yet it currently faces a critical dilemma: researchers must choose between sparse outcome-based rewards, which provide insufficient logical guidance, or expensive neural Process Reward Models (PRMs) for dense signals. We resolve this by introducing SPEAR (Symbolic Process Evaluation and Alignment Reward), a training-free and plug-and-play process reward method for sequence-level on-policy distillation. SPEAR projects natural-language reasoning traces into domain-adaptive symbolic milestones, providing an efficient proxy for process-level reasoning alignment. By utilizing the longest common subsequence (LCS) to align student explorations with teacher milestones, SPEAR provides a dense, order-aware reward signal that aligns logical consistency without the need for an external neural verifier. Our experiments across math, science, and commonsense reasoning tasks demonstrate that SPEAR effectively bridges the reasoning gap between student and teacher models via sequence-level distillation with efficient dense process rewards. Our code and data are available at: https://github.com/zhuochunli/SPEAR.
Read the full abstract

All authors Zhuochun Li, Yuelyu Ji, Yiming Zeng, Daqing He