Robotics
PhasePlan: Ordered Future-Phase Planning for Robot Brain Models
Xiaoyu Yang, Yafei Zhang, Wensheng Li et al. (+2 more)
Robot brain models integrate
vision,
language, and
robot state to generate actions for complex manipulation tasks. Most predict fixed-length action chunks that may span multiple task phases. This can obscure phase transitions and favor frequent action patterns, compromising action timing in dynamic environments. We propose \method, an ordered future-phase planning method for
robot brain models. From current
multimodal observations, it predicts the task phase at each future action position. The resulting planning representations condition the corresponding actions, preserving temporal
alignment between task progress and action generation. Training first learns the planner, then freezes it during action-model adaptation to maintain stable phase representations. We instantiate \method on pretrained $\pi_{0.5}$ and AcrossWAM1.0
robot brain models. Detailed quantitative evaluation uses the $\pi_{0.5}$ implementation. On conveyor-belt manipulation, \method reduces offline joint-action error by approximately 22.5% relative to the original $\pi_{0.5}$ model. It also improves phase-transition modeling and cross-phase action prediction. These results demonstrate the value of ordered future-phase planning for continuous action generation.
Read the full abstract ▾
All authors Xiaoyu Yang, Yafei Zhang, Wensheng Li, Qing Zhan, Nan Wu
Robotics
TACTIC: Temporal and Context-Aware LLM Tactical Planning for Roadside LiDAR Attacks
Yiming Gao and Shaocheng Luo
Physical
LiDAR attacks are often evaluated using fixed primitives and manually selected parameters, despite their strong dependence on surrounding traffic. We present TACTIC, a scene-aware framework that uses a
multimodal large
language model (MLLM) to coordinate state-adaptive roadside
LiDAR attacks. Under a gray-box threat model, TACTIC relies only on an attacker-operated roadside perception stack, without accessing the victim
LiDAR's native point clouds or internal processing. Local perception provides metric vehicle states, while the MLLM combines these measurements with roadside imagery to infer relational traffic context and construct a semantic scene graph. Based on this representation, TACTIC selects and configures two complementary primitives: push-away, which shifts the perceived range of a lead vehicle, and phantom-obstacle braking, which triggers emergency braking through obstacle injection. Measured traffic states and empirically calibrated constraints ground the generated tactics in physically feasible operating regions. To accommodate MLLM latency, TACTIC overlaps
reasoning and execution asynchronously while high-rate local perception detects scene changes and triggers replanning. Across 280 randomized CARLA trials, the full policy achieves a 100% collision rate, compared with 35% for a fixed rule, 60% for random selection, and 75% for a restricted LLM using mode selection with default parameters. Joint physical-and-image input achieves 100% success, versus 65% with physical measurements alone and 75% with imagery alone, while asynchronous $\Delta$ refresh reduces scene-mutation response from 7.4 s to 2.0 s. These results show that scene-dependent tactical planning can expose context-sensitive
LiDAR failure modes that fixed attack policies may miss.
Read the full abstract ▾
Systems and Control
On the Input Sensitivity of METANET Models and the Robustness of Dynamic Calibration
Cameron Hickert, Shreyaa Raghavan and Cathy Wu
While traffic modeling and control rely upon effective calibration of macroscopic traffic simulation models like METANET, recent empirical work shows these models can exhibit severe sensitivity to input noise. This paper explains this phenomenon through a string-stability analysis, demonstrating that a calibrated METANET model can amplify small additive perturbations to boundary conditions along the corridor, causing the simulated state to diverge from the nominal baseline. The input sensitivity in off-nominal cases may compromise the model's capabilities for counterfactual analysis, which is required for use cases of interest like the design of large-scale variable speed limits. To address these issues, this work demonstrates how a dynamic, time-varying approach to calibrate model parameters can achieve robustness and improved accuracy of the resulting macrosimulation. We show analytically that under stated regularity and perturbation assumptions, dynamic calibration achieves a tighter cost deviation bound than static calibration, and we validate this behavior in highway environments with synthetic and real-world data. Ultimately, this can enable more trustworthy traffic simulation and more effective evaluation of control strategies.
Read the full abstract ▾
Machine Learning
What Limits Recursive Reasoning Models: Optimization, Architecture and Test-Time Scaling
Yuliana Shakhvalieva, Dmitrii Kharchev, Viacheslav Bezrukov et al. (+4 more)
Recursive
reasoning models apply a small shared
Transformer block many times to refine a latent state. This gives them large effective depth with few parameters and makes them strong on algorithmic tasks. Such compact solvers are natural candidates for tools that an LLM can call on narrow algorithmic subproblems. However, existing models such as HRM, TRM and URM differ in architecture, gradient propagation and training procedure simultaneously. This makes it hard to tell what drives their performance, and their optimization is still poorly understood and often unstable. In this work we address both of these gaps. First, we study these questions under a unified experimental pipeline spanning six algorithmic domains. Individual controlled ablations are performed on representative domains, while the resulting recipe is evaluated across the full suite. The study reveals a surprisingly simple recipe for stable and generalizable recursive
reasoning: an intermediate gradient horizon, large physical batches and controlled updates of the recurrent state. An explicit hierarchical architecture is not needed. Second, we combine these findings into a stable 13.6M-parameter model that achieves the strongest overall performance among the evaluated recursive baselines, with particularly large gains on out-of-distribution generalization. It raises Arithmetic OOD accuracy to 71.2%, from 36.2% for the strongest baseline, while reaching 98.41% on Sudoku and 59.5% pass@2 on ARC-AGI-1. Our results show that, within the recursive architectures studied here, performance depends strongly on how recurrence is optimized and stabilized. More broadly, it shows how AI systems can be improved by optimizing their components one at a time.
Read the full abstract ▾
All authors Yuliana Shakhvalieva, Dmitrii Kharchev, Viacheslav Bezrukov, Inessa Fedorova, Dmitry Bocharov, Ivan Oseledets, Valerii Ternovskii
Artificial Intelligence
AIMS: An Agentic AI Framework for Sim-to-Real Multi-Modal ISAC
Yijie Bian, Kai Zhang, Wei Guo et al. (+4 more)
Multi-modal integrated sensing and communication (
ISAC) enables environmental perception and reliable connectivity for intelligent wireless networks. Data-driven multi-modal
ISAC models depend heavily on annotated real-world data to learn relationships across sensing and wireless observations, thereby constraining scalable deployment. Although synthetic data generation reduces the burden, adapting existing simulation pipelines to a target deployment requires consistent scene, sensing, wireless, and learning configurations, while mismatches among these coupled components impair sim-to-real transferability. To address the challenge, we propose an agentic artificial intelligence (AI) framework for sim-to-real multi-modal
ISAC, named AIMS. Given a natural-
language deployment request specifying the target task, deployment conditions, and real-data budget, AIMS derives a deployment-specific sim-to-real configuration and coordinates its execution to produce a deployment-specific task model. A two-
agent architecture coordinates scene construction with task learning. A scene construction
agent generates geographically grounded, synchronized sensing and wireless records from shared physical states, while a scene understanding
agent configures task-relevant modalities and mixture-of-experts (MoE) learning for zero-shot inference or few-shot adaptation. Structured domain knowledge guides dependency-aware planning, while validation evidence supports feedback-driven revision of affected decisions. Experiments on the real-world DeepSense 6G
dataset demonstrate improved vehicle detection and beam prediction over the considered simulation and fusion baselines. A separate orchestration
benchmark evaluates task interpretation, dependency
reasoning, and feedback-driven replanning across diverse deployment requests, showing improved plan correctness with structured domain knowledge and validation feedback.
Read the full abstract ▾
All authors Yijie Bian, Kai Zhang, Wei Guo, Zixin Wang, Shenghui Song, Jun Zhang, Khaled B. Letaief
Human-Computer Interaction
Diptych: Scoped, AI-Interpreted Comparison for Reference Listening in Music Production
Chongjun Zhong, Abhinaba Roy, Archishman Ghosh et al. (+2 more)
Reference listening is a common strategy in music production, but current comparison tools often obscure a key human judgment: deciding what should be compared. We present Diptych, an AI-assisted system that lets users define comparison scope across whole tracks or independently selected segments, while inspecting structured audio features and scope-specific AI interpretations. We evaluated Diptych in a within-participants study with 12 musicians, complemented by source-blinded ratings from four expert listeners. Participants used the system to surface additional differences, nine of ten of which received at least partial expert support, and reported good usability and greater clarity about possible next steps. These findings suggest that AI support for creative comparison should prioritize user-defined scope, inspectable evidence, and actionable guidance, while avoiding authoritative judgments that exceed what the evidence can support.
Read the full abstract ▾
All authors Chongjun Zhong, Abhinaba Roy, Archishman Ghosh, Kejun Zhang, Dorien Herremans
Computer Vision
Reconstructing the Dynamic World: A Representation-Centric View of 4D Scene Reconstruction
Ziren Gong, Guo Chen, Yongjia Li et al. (+13 more)
4D scene reconstruction aims to recover the evolving geometry, appearance, and motion of dynamic environments from visual observations. Despite substantial progress in neural scene representations, reconstructing dynamic scenes remains challenging due to non-rigid motion, occlusions, temporal inconsistencies, and the trade-offs between reconstruction fidelity and computational efficiency. Recent advances in Neural Radiance Fields (NeRF) and 3D Gaussian Splatting (3DGS) have introduced diverse approaches to representing and reconstructing dynamic scenes, yet their relationships, underlying design choices, and evaluation protocols remain fragmented. In this paper, we present a unified perspective on 4D scene reconstruction, organizing existing methods around their scene representations, temporal modeling strategies, reconstruction pipelines, and optimization objectives. Through this framework, we examine how different design choices affect
geometric fidelity, appearance consistency, motion representation, and computational efficiency. We further consolidate commonly used
datasets and evaluation metrics, identify limitations in current experimental practices, and discuss open challenges in reconstructing complex, dynamic real-world environments. By connecting methodological developments with their underlying assumptions and evaluation evidence, this work provides a structured foundation for understanding existing approaches and identifying future research directions. An evolving collection of relevant papers and resources is available at https://github.com/ZiyangYan/Awesome-4D-Scene-Reconstruction.
Read the full abstract ▾
All authors Ziren Gong, Guo Chen, Yongjia Li, Yihua Shao, Fabio Tosi, Stefano Mattoccia, Matteo Poggi, Hao Tang, Fei Ma, Shuyan Li, Ziyang Yan, Nicu Sebe, Ling Shao, Jianfei Cai, Qi Tian, Ming-Hsuan Yang
Artificial Intelligence
Better Deck or Different Judge? Evaluating Agentic Harness Gains in Corporate and Investment Banking
Ludovic Gibert, Matis Despujols and Andre-Louis Rochet
Corporate and investment banking teams use presentations to support credit decisions and advise clients on financing and transactions. Producing these decks requires reconciling financial data, tracing sources and turning analysis into a recommendation. We retrospectively study the development of an agentic harness combining a 27B
language model, financial calculations, narrative templates and validation checks. LLM judges guide engineering changes and assess the resulting decks, raising the question of whether higher scores reflect better documents or changes in grading. In shared-session text-only grading with template markers removed, five judges score the complete system 20.4 to 33.6 points out of 95 above the same model generating directly from a short
prompt. Every judge scores the system higher on all seventeen development deliverables. Margins against direct Opus generation from a short
prompt range from -4.7 to +0.8 points. Judges agree on broad progress across development rounds but agree less on final-deck rankings than on pooled scores. Repeated grading also shifts scores on unchanged decks, making small improvements difficult to distinguish from judge variability.
Read the full abstract ▾
Software Engineering
Learning When and How to Intervene: A Hindsight-Distilled Sentinel for Coding Agents
Jiangrui Zhao, Chenglong Li, Meng Zhang et al. (+1 more)
Coding
agents solve repository-level tasks through sequences of actions, where a single erroneous action can misdirect subsequent decisions and increase recovery costs. Existing approaches use execution feedback for recovery or specialized checks to block errors, but deciding before execution whether intervention will benefit eventual task completion remains challenging. To address this challenge, we propose HiSentinel, a hindsight-
distillation framework that trains lightweight 0.6B and 1.7B sentinels to select pre-execution interventions aimed at improving task completion rather than correcting every imperfect action. A privileged teacher uses recorded execution outcomes as evidence for intervention judgments, which are distilled into a causal student that receives only the pre-action context and proposed action. Beyond identifying whether and when to intervene, the sentinel must also provide actionable feedback that helps the coding
agent recover or obtain necessary human input. To support these capabilities, we introduce SWE-Intervene, an action-level
dataset constructed from software-engineering trajectories that annotates whether an action should be allowed, autonomously redirected, or paused for human assistance, together with corresponding intervention feedback. Across SWE-bench Verified Mini and Ask or Assume, HiSentinel consistently improves task completion across Sentinel scales and coding-
agent families, with gains of up to 14% and 10%, respectively, while maintaining competitive
token consumption. These results demonstrate that lightweight pre-execution intervention can effectively prevent error propagation and improve the reliability of autonomous coding
agents.
Read the full abstract ▾
All authors Jiangrui Zhao, Chenglong Li, Meng Zhang, Xiaoting Du
Artificial Intelligence
Coverage Before Control: Route-Instruction Grounding and Steering for Controllable Retrosynthesis
Xuemin Chen, Xiaozhuang Song, Xinjian Zhao et al. (+2 more)
Single-step retrosynthesis models are commonly evaluated by their ability to recover recorded reactions. In practice, chemists may need to choose among several precursor sets for the same product, for example to preserve a particular motif. Recovering a recorded answer alone does not establish this ability to follow a preference. Satisfying such requests requires both coverage of relevant alternatives and control over which alternatives are favored. We introduce Route-Instruction Grounding and Steering (RIGS), a two-stage framework for instruction-conditioned retrosynthesis. Stage A trains a
language projector, teaching it which alternatives an instruction favors or discourages. Stage B uses the projector learned in Stage A to steer a frozen generative model through lightweight residual adapters. We construct nested one-to-many training supports by pairing each product with increasing numbers of candidate precursor sets. Extensive experiments demonstrate that broader support helps the model generate a wider range of alternatives, and RIGS can learn to guide generation according to instructions. The relationship between coverage and control is consistent across model scales but non-monotone.
Read the full abstract ▾
All authors Xuemin Chen, Xiaozhuang Song, Xinjian Zhao, Yaoyao Xu, Tianshu Yu
Computer Vision
Learning to Reason with Compressed Context: Ground-Truth-Free Adaptation of OmniLLMs via Self-Distillation
Jianghao Wang, Ke Meng, Jian Li et al. (+7 more)
Omni-modal large
language models (OmniLLMs) enable unified audio-
video understanding, but their long
multimodal token sequences make deployment computationally expensive.
Token compression reduces this cost, yet aggressive compression often lowers accuracy. Existing works predominantly focus on designing better compression mechanisms; however, adapting the underlying
language model to reason effectively over the remaining compressed context remains under-explored. To address this, we propose CAFD (Compressed-Context Adaptation via Full-Context
Distillation), a ground-truth-free self-
distillation framework that adapts OmniLLMs to fixed compression pipelines without requiring reference answers, rationales, or correctness rewards. CAFD leverages the full-
token view of the same
multimodal sample as a source of privileged information: a full-context self-teacher provides soft target supervision to a compressed-context student along the student's on-policy trajectory. Evaluated on Qwen2.5-Omni-7B across five audio-
video benchmarks, five compression pipelines, and five deployment budgets, CAFD demonstrates consistent gains, improving 120 out of 125 conditions with an average accuracy boost of 1.44 points and recovering 26.9% of the accuracy gap on average. These results demonstrate that the proposed ground-truth-free adaptation offers an effective and practical route to improving the accuracy-efficiency trade-off in deployed OmniLLMs.
Read the full abstract ▾
All authors Jianghao Wang, Ke Meng, Jian Li, Chi Cheng, Longyu Qi, Liyin Liang, Yifeng Qian, Chunbo Lai, Yutian Lin, Zeyu Wang
Robotics
Magnetic based In-situ Self 3D Pose Estimation for a Modular Soft Tendon-Driven Continuum Robot via IMU-Fusion
Zheng Cao, Guo Ning Sue, Xiangyun Bu et al. (+3 more)
Continuum
robots are well suited for gentle manipulation because of their inherent compliance and ability to adapt to complex environments. However, their continuously deformable structure makes accurate configuration estimation challenging, particularly when external
vision systems are unavailable or obstructed. In this work, we present an embedded pose sensing framework that combines inertial measurement units (IMUs) and active
magnetic fields to estimate the
robot configuration without relying on external cameras. The angular measurements from the IMU and
magnetic-field references are fused to improve local orientation estimation and reduce accumulated orientation error during operation. This pose sensing scheme achieves an update rate of 16.7~Hz, allowing real-time feedback. The proposed system is experimentally validated through closed-loop control, where the estimated
robot configuration is used to maintain the end-effector at a desired position while interacting with an object. These results demonstrate the potential of distributed
magnetic–inertial sensing for real-time pose estimation and closed-loop control of continuum
robots.
Read the full abstract ▾
All authors Zheng Cao, Guo Ning Sue, Xiangyun Bu, David Quinn, Junzhe Hu, Carmel Majidi
Numerical Analysis · Cross-listed from math.PR
Computing Extinction Barriers in a Quorum-Sensing Reaction Network
Mario Ayala and Johannes Zimmer
We introduce a simple reaction-network model of a quorum-sensing population that couples the
cell density $x$ to the signal density $w$. In the four-channel
cell-signal network studied here, scaling signal production and removal by the same factor $r$ leaves all deterministic
equilibria, and their stability types, unchanged. Nevertheless, we show that $r$ shifts the quasipotential barrier $\Delta V(r)$ for rare transitions towards extinction, and thus, under metastable exit assumptions, the mean time to reach a fixed neighbourhood of the extinction state on the exponential scale $e^{N\Delta V(r)}$. We compute the barrier by minimization of the path action with the signal retained as a fluctuating coordinate, and we check it against mean extinction times obtained without the action solver, i.e., from the backward equation of the jump process on a truncated
lattice and from exact
stochastic simulation. Over a range of $r$, the minimum-action barriers satisfy $\Delta V(r)=\Delta V_\infty+O(1/r)$, where $\Delta V_\infty$ is obtained by eliminating the signal first, and at $r=1$ the barrier is 55% larger than $\Delta V_\infty$. A formal first-order expansion attributes this increase to the lag of the signal, of which the fluctuations of the signal remove a quarter. The saddle barrier $\Delta V(r)$ is also the least action needed to enter the basin of extinction, whereas the density threshold can be crossed more cheaply, by paths which keep the signal high and are usually followed by recovery. The mean extinction times of the backward equation, which include failed attempts, have a prefactor which tends to a constant, and they grow at a rate within $10^{-5}$ of $\Delta V(r)$ at every tested rate.
Read the full abstract ▾
Computation and Language
LEAP: Learned Block-wise Evidence Retrieval for Long Audio-Video Perception
Juyi Lin, Zhiqiang Lao, Jiali Cui et al. (+9 more)
Hour-scale audio-visual question answering is constrained by a context dilemma: dense whole-recording encoding rapidly exhausts context limits, whereas uniform temporal compression severely dilutes fine-grained acoustic and visual evidence. We introduce LEAP, a framework where the model retrieves its own evidence without placing the whole recording in one context. LEAP divides a recording into fixed-duration blocks, applying a lightweight localization pass to each block to score short candidate windows. The highest-ranked windows are pooled and re-encoded in a single bounded answer pass. Consequently, the answer input and peak context remain independent of the recording duration. By decoupling evidence localization from
reasoning, our framework can localize candidate temporal windows over pre-computed transcripts without decoding media frames, while preserving fine-grained visual and non-
speech evidence by routing the final answering pass over raw audio-visual streams. LEAP trains both stages: a localization LoRA improves the selected windows, and an answer LoRA improves the answers read from the same windows. The block grid natively supports causal queries, enabling LEAP to support streaming inference without streaming-specific training. Across several AVQA
benchmarks, LEAP improves over the Qwen3-Omni-30B-A3B baseline by 4.5-16.8%, and transfers to a second omni-modal backbone, MiniCPM-o 4.5, surpassing its published results by 3.1-13.0%.
Read the full abstract ▾
All authors Juyi Lin, Zhiqiang Lao, Jiali Cui, Lin Zhao, Pu Zhao, Dichang Zhang, Arman Akbari, Yu Qi, Xinru Jiang, Yanzhi Wang, Heather Yu, Liang Peng
Systems and Control
Globally Certified Invariant-Ellipsoid Control from Data
Ngoc Tuan Dinh, Egor Dogadin and Alexey Peregudin
This letter develops a data-based method for designing state feedback for a discrete-time linear system under bounded disturbances. For each admissible feedback gain and scalar design parameter, a Lyapunov equation determines an invariant ellipsoid: a region that the state cannot leave under the permitted disturbances. We optimise the gain and parameter to minimise a trace-based measure of the resulting output enclosure. At each parameter value, value iteration gives a lower cost bound, while separate controller evaluation gives an achievable upper bound. These bounds also let us assess whole parameter intervals without evaluating every point. We prove that the search stops after finitely many evaluations with a stabilising state-feedback gain whose objective is within any prescribed absolute tolerance of the infimum over the chosen ellipsoid family. Neither an initially stabilising gain nor attainment of the infimum is assumed. The guarantee assumes exact arithmetic and a sufficiently informative batch of exact measurements, including disturbances during data collection; the resulting feedback uses only the state. A position–velocity example illustrates the bounds, controller checks, and computational cost.
Read the full abstract ▾
Machine Learning
Reliability-Aware Checkpoint Selection for Domain Generalization
Jinshi Liu, Jiahao Li, Pan Liu et al. (+5 more)
Checkpoint selection in domain generalization often relies on source-validation accuracy, yet the selected checkpoint need not provide reliable probabilities on unseen target domains. Source-target distribution shifts can alter accuracy rankings, while accuracy alone does not measure predictive probability quality. We identify an empirical selection opportunity within fixed training trajectories: reselecting among checkpoints with near-optimal source accuracy can improve mean target probability quality with small observed changes in mean target accuracy. We study accuracy-constrained reliability selection (AC), which retains checkpoints within a tolerance of the best source-validation accuracy and ranks them by source reliability. Our reference rule aggregates within-set normalized negative log-likelihood (NLL) and class-wise calibration error (CwECE) using $D_\infty$. AC uses no target data and requires neither additional training nor weight averaging. We evaluate five domain generalization training
algorithms on three
benchmarks, using PACS to develop the objectives and a 0.5-percentage-point tolerance. In exploratory aggregation comparisons on 360 OfficeHome and TerraIncognita runs, the reference rule reduces mean target soft-bin squared-gap ECE and CwECE by 0.240% and 0.182%, respectively, and NLL by 0.030 relative to Source-Acc. Mean target accuracy changes by +0.213 percentage points. These results identify opportunities for reliability-aware reselection, while the additional benefit of joint over single-objective ranking remains unresolved.
Read the full abstract ▾
All authors Jinshi Liu, Jiahao Li, Pan Liu, Yanfeng Li, Rui Qian, Zhao Tong, Yue Sun, Tao Tan
Artificial Intelligence
ConflictGuide: AutoResearch Improves When Competing Behaviors Are Made Visible
Binqian Xu, Qiran Zou, Xiangbo Shu et al. (+1 more)
When designing machine learning models, desirable properties are often in tension: improving one behavior can impair another, so task progress can depend on alleviating the conflict. LLM-based AutoResearch systems, which iteratively edit model code and retain edits based on scalar task-performance feedback, have largely ignored this trade-off. We find that scalar feedback supports broad exploration early in search, but it does not reveal how edits affect competing behaviors. In matched-budget experiments, introducing competing-behavior feedback as task gains diminish increases the share of proposals that improve both behaviors and sustains progress beyond scalar-only plateaus. Obtaining this feedback for a given model requires identifying its competing behaviors and designing probes to measure them. To make competing-behavior feedback actionable, we introduce ConflictGuide. Its reusable ConflictGuide-Skill combines a literature-grounded taxonomy with model-specific evidence to identify competing behaviors and specify probes for a code
agent to implement as metrics.
Evolution proceeds in two stages: Stage I explores with task feedback; Stage II uses probe feedback to steer proposals toward conflict alleviation and retains marginal-gain edits only when probes indicate sufficient alleviation. Across five diverse model families, ConflictGuide reduces task and conflict-related errors by up to 28% and 14%, respectively, relative to scalar-only AutoResearch, with gains extending to other code
agents.
Read the full abstract ▾
All authors Binqian Xu, Qiran Zou, Xiangbo Shu, Dianbo Liu
Machine Learning
RoPE at the End of Its Rope? Theory, Diagnosis, and Mitigation of Long-Context Failures
Yuyang Wu, Yufeng Du and Hao Peng
Long-context failures of RoPE-based
language models can arise from RoPE's intrinsic tradeoff between maintaining stable
token preferences and distinguishing nearby positions. Determining which weakness to address, and how, requires a more precise characterization of RoPE's behavior in trained models across context lengths. We address a key limitation of prior theory by allowing unequal query-key scales across RoPE frequencies, which aligns well with practical empirical observations. Our theory makes both vulnerabilities measurable for individual heads and inputs, and quantifies how high-frequency components support positional sensitivity while potentially disrupting semantic stability. We also derive a theoretical context-length bound beyond which, under specified conditions, a fixed
attention-score comparison cannot jointly avoid semantic reversal and positional insensitivity. Guided by our fresh theoretical insights, we introduce RoPE Profiler, a lightweight, plug-and-play diagnostic toolkit that augments existing evaluations with zero additional forward passes by reusing cached query and key activations. Reusing activations collected during evaluation, the toolkit incurs little overhead. It supplements standard
benchmark scores with two diagnostic scores that reveal semantic and positional weaknesses and help users prioritize which aspect to address. Crucially, our evaluations across 49 long-context task settings reveal a distinct pattern where
reasoning tasks predominantly suffer from semantic reversal, whereas
retrieval tasks are primarily vulnerable to positional insensitivity. Guided by our theory and diagnostic profiles, targeted high-frequency rescaling achieves immediate gains without additional training, improving task accuracy by up to 20 percentage points on Qwen3-8B and 25 percentage points on Llama-3.1-8B-Instruct.
Read the full abstract ▾
Computation and Language
AdaGEPA: Adaptive Feedback Allocation for Reflective Prompt Optimization
Junyang Chen, Zecheng Wang and Jingbang Chen
Prompt optimization improves the performance of
language-model systems on downstream tasks by refining their
prompts. Classical methods evaluate
prompts on task examples and use the resulting feedback to guide
prompt revisions through reflection. However, when feedback selection does not account for the
prompt's weaknesses, these revisions may improve performance on selected examples without yielding broader task improvements. To address this issue, we propose AdaGEPA, an adaptive feedback-allocation method that uses the
prompt's performance and task structure to select examples for the next
prompt revision. Our method replaces at most one example in each feedback minibatch to target an identified weakness while preserving the remaining feedback context. Across our main experiments on six downstream
benchmarks, AdaGEPA achieves higher mean validation scores than non-adaptive feedback selection under matched rollout budgets. AdaGEPA also finds high-performing
prompts earlier across several tasks. In the initial Schema-Guided Dialogue (SGD) study, its half-budget
prompts outperform the non-adaptive baseline's full-budget
prompts in joint goal accuracy on new dialogues from services seen and unseen during search. Overall, our findings highlight the potential of adaptive feedback allocation to improve both the effectiveness and rollout-budget efficiency of reflective
prompt optimization.
Read the full abstract ▾
Computer Vision
Super-Resolving Unseen Hyperspectral Sensors at Any Scale via Spatial Operators
Ji-Xuan He, Guohang Zhuang, Bo Junge et al. (+6 more)
Achieving cross-sensor generalization and arbitrary-scale reconstruction with a single model remains challenging in hyperspectral super-resolution (HSR). Although recent methods support arbitrary-scale reconstruction, applying them to new sensors or scales beyond the training range often requires additional data and computation to maintain reconstruction quality. To address these challenges, we propose OmniHSR, which predicts band-shared spatial operators rather than spectral values. Cross-Spectral Mapping (CSM) resamples inputs with any number of bands to fixed reference positions and predicts local operators with Gaussian supports. Continuous Operator-Field Reconstruction (COFR) composes these operators into a continuous field and applies them to all original bands for arbitrary-scale reconstruction. Experiments demonstrate that operator prediction outperforms direct spectral-value prediction on all seven
datasets. Trained solely on ARAD with only 0.538M parameters, OmniHSR outperforms all directly transferred baselines on six unseen
datasets without target-domain training data or adaptation. Across twelve upsampling factors from $\times2$ to $\times48$, it improves average PSNR on Pavia U and Chikusei by 0.55 dB over the strongest baseline. It also surpasses baselines trained from scratch or adapted on the target sensor and achieves up to $36\times$ faster inference. Our code will be publicly released soon.
Read the full abstract ▾
All authors Ji-Xuan He, Guohang Zhuang, Bo Junge, Tingyi Li, Lingchen, Miaomiao Cai, Yanan Qiao, Xiujin Liu, Junfeng Fang
Computer Vision
CoVisco: Codec-Native Vision Encoder with Native Token Compression for Unified Image-Video Understanding
Yulong Liu, Xiaotian Han, Junyuan Shang et al. (+6 more)
Vision-
language models face a fundamental scaling bottleneck: the number of visual
tokens grows with both temporal duration and spatial resolution, making long-
video understanding expensive for the
vision encoder and the
language model. Existing methods often compress visual
tokens after dense encoding, creating a mismatch between the representation used during training and the compact interface required at deployment. We present CoVisco, a codec-native
vision encoder with native
token compression for unified image-
video understanding. By combining codec-native input support with segmented
attention, CoVisco can encode long visual inputs in a single forward pass without forming dense patch-to-patch interactions across all frames. Each temporal segment is equipped with learnable abstract
tokens that learn a compact segment-level representation, while fine-grained patch
tokens remain available throughout the encoder. Alternating intra-segment and abstract-communication layers preserve
video-level context through the abstract-
token channel. A lightweight selector further exposes either abstract
tokens alone or abstract
tokens augmented with a runtime-selected subset of patch
tokens, yielding a compact visual interface that reduces the visual context and prefill burden of downstream MLLMs while retaining fine-grained evidence when needed. Pretrained with contrastive objectives on 565M image–text pairs and 6.4M
videos, CoVisco shows competitive performance on
video-oriented
embedding and
multimodal understanding
benchmarks. In the evaluated four-segment, 64-frame setting, abstract-only inference uses only 400 visual
tokens while achieving
video-understanding performance close to, and on some
benchmarks exceeding, OneVision-Encoder. Selected patch
tokens further improve fine-grained
video reasoning. Project URL: https://github.com/ernie-research/CoVisco.git
Read the full abstract ▾
All authors Yulong Liu, Xiaotian Han, Junyuan Shang, Yuchen Ding, Zhenyu Zhang, Shuohuan Wang, Guibo Zhu, Sirui Han, Dianhai Yu
Computer Vision
Yanshu Li, Jiaqian Li, Canran Xiao et al. (+3 more)
Large
vision-
language models (LVLMs) exhibit strong
multimodal in-context learning (ICL) capabilities, yet this ability degrades substantially as model size decreases. Knowledge
distillation offers a natural way to bridge this gap, but existing methods primarily align output distributions or hidden representations directly. Such
alignment teaches the student what the teacher predicts without revealing which evidence in the complex context causally supports that prediction. Consequently, a student can imitate the teacher's answer while continuing to rely on
language priors,
prompt structure, or other spurious cues. To address this limitation, we introduce
Multimodal Causal
Distillation (MCD), a
distillation framework that transfers how a strong teacher uses
multimodal evidence during ICL. MCD uses structure-preserving
token interventions to identify and verify causal evidence, then transfers how the teacher responds when that evidence is retained or removed. This design connects
distillation to the causal patterns by which the model uses contextual evidence during
multimodal ICL. Experiments across three LVLM families and seven
benchmarks show that MCD improves student performance by 7.23 points on average and outperforms vanilla
distillation by 4.68 points, while further analyses confirm the generalizability of these gains.
Read the full abstract ▾
All authors Yanshu Li, Jiaqian Li, Canran Xiao, Xi Xiao, Tianyang Wang, Yongtai Liu
Computational Complexity · Cross-listed from quant-ph
Verifiable quantum advantage based on polynomials with planted structures
Markus Bläser, Michael Gullans, Dominik Hangleiter et al. (+2 more)
A central question in the theory of
quantum advantage is whether there are
quantum advantage protocols with similar resource requirements as random
circuit sampling that are also verifiable just from the classical outputs of the
quantum computation. Here, we develop the idea of simulation secrets for verifiable advantage. A verifier can use a simulation secret to evaluate a cross-
entropy test faster than it would take a classical adversary to pass the test. We instantiate this idea using IQP
circuits described by cubic
polynomials with planted independent spaces. These correspond to the largest independent set in the orbit of a
polynomial under the general linear group and yield a low-rank stabilizer decomposition of the corresponding state. We
conjecture that large independent spaces are invisible to a computationally bounded adversary, and therefore they cannot exploit them to pass the protocol. A second
conjecture regards the fine-grained complexity of producing samples that pass the cross-
entropy test for uniformly random
polynomials. Under these
conjectures, our scheme results in a
polynomial gap between the verification time and the time a classical adversary would need to pass the protocol—both are exponential. It has a potential application to generating classically certifiable randomness, since the output distributions have high min-
entropy. We estimate that the planted
polynomial scheme is implementable using 100 logical
qubits at logical error rates around $10^{-6}$.
Read the full abstract ▾
All authors Markus Bläser, Michael Gullans, Dominik Hangleiter, Yuxuan Liu, Youming Qiao
Discrete Mathematics · Cross-listed from math.CO
Log-concavity and Approximate Counting for Totally Unimodular Polytopes
Jonathan Leake and Maryam Mohammadi Yekta
We present a new lower bound on the number of
lattice points of all totally unimodular polytopes, generalizing previous lower bounds on contingency tables, integer flows, and beyond. Our bound is based on the Gurvits capacity convex optimization problem, and thus our result implies an efficient deterministic
algorithm for approximate counting of the
lattice points up to an explicit exponential factor. We achieve our bounds by showing that the associated generating
polynomials fit into a new general class of log-concave
polynomials called VLC ("variable-wise log-concavity”). This also implies a
conjecture of Ferroni and Higashitani on the evaluations of the Ehrhart
polynomials of unimodular polytopes. The essential ingredient for these results is the resolution of Barvinok's log-concavity
conjecture for contingency tables on lines, which was proven using ChatGPT 6 Astra. We
conjecture a generalization of Barvinok's
conjecture, which we believe will lead to stronger and more general bounds.
Read the full abstract ▾
Computer Vision
Haoxiang Shi, Zaijing Li, Muhe Ding et al. (+3 more)
Vision-
Language Navigation (VLN) requires
embodied agents to generate actions based on instructions and observations. General-purpose
multimodal agents offer a promising basis for this task, but selecting plausible local actions does not ensure that execution remains consistent with the intended route, particularly in long-horizon tasks. Moreover, the accumulated interaction history increases the input required for subsequent decisions, resulting in a significant inference overhead. To this end, we introduce \method, an Agentic VLN framework that includes a Goal
Agent that sets adaptive goals for local actions, a Verify
Agent that dynamically verifies whether a goal has been completed, a
Memory Agent for
multimodal context compression, and a Visuomotor
Agent to execute adaptive goals. Specifically, the Goal
Agent formulates adaptive goals based on the instruction, current observation, and execution history. Then the Visuomotor
Agent executes
navigation actions to achieve each goal, while the Verify
Agent uses a goal-specific verification question to dynamically assess whether the observed outcomes satisfy the intended completion condition. Verified goal completion then marks a boundary for the
Memory Agent to compress the corresponding
multimodal interaction history while preserving information needed for subsequent
navigation. We evaluate
navigation on R2R-CE and RxR-CE, examine framework variants across three model backbones, and study context
evolution during execution. For Real-World evaluation, \method achieves 83.3% success and 1.51\,m
navigation error across eight challenging routes evaluated three times each.
Read the full abstract ▾
All authors Haoxiang Shi, Zaijing Li, Muhe Ding, Xiang Deng, Yaowei Wang, Liqiang Nie
Artificial Intelligence · Cross-listed from math-ph
Cluster Attention Neural Operators for Solving Parametric Partial Differential Equations
Ming Zhong, Antonio Colanera, Gianluigi Rozza et al. (+1 more)
Traditional simulations of parametric partial differential equations (PDEs) rely on repetitive computations for each parameter, which makes high-fidelity design impractical. Neural operators address this issue by learning solution operators, accelerating parameter-space mapping by orders of magnitude. Recent
Transformer-based neural operators attempt to capture global dependencies, but often at the cost of quadratic
attention complexity. Transolver resolves this problem by projecting physical states into a reduced slice space for
attention computation. Although fast, this projection sacrifices fine spatial information. Moreover, by operating in this reduced space with shared weights across
attention heads, it may constrain the model's flexibility, thereby limiting its capacity to capture complex phenomena. To address these issues, we propose the Cluster
Attention Neural Operator (CANO), which reformulates
attention via a novel cross-
attention mechanism that dynamically clusters queries while preserving full-resolution keys and values. This avoids slice compression loss and removes weight-sharing limits. At the same time, the model remains fast without losing global interactions. Empirically, CANO achieves state-of-the-art performance across canonical PDE
benchmarks, covering fluid and solid dynamics (e.g., Navier-Stokes, Airfoil, Plasticity), irregular unstructured geometries (e.g., Pipe
Turbulence, Composites), and long-term temporal rollouts. Across solid deformation and turbulent flow
benchmarks, CANO achieves lower errors than baselines and exhibits strong
geometric adaptability and temporal consistency.
Read the full abstract ▾
All authors Ming Zhong, Antonio Colanera, Gianluigi Rozza, Zhenya Yan
Computation and Language
The Concrete-Arbitrary Gap: Kinship Reasoning in LLMs Is Not Indifferent to Presentation
Thomas Pashby
We test whether large
language models solve formally matched kinship problems equally well when relations are expressed in familiar vocabulary or by explicitly defined nonce predicates. Across 500 paired graphs, concrete accuracy exceeds arbitrary accuracy by 35.6 percentage points in local Qwen3.8-27B, 26.6 in Gemma 4 26B-A4B, 12.0 in Gemma 4 31B, and 5.4 in Qwen3.8-Max. All four paired gaps are statistically resolved.
Reasoning budgets and
prompt-
language interventions can substantially reduce the difference, showing that it is modifiable rather than a fixed incapacity. The minimal conclusion is behavioral: on these tasks, the models' manifested relational competence is not indifferent to presentation. Explicit definitions provide the formal relations but do not make nonce predicates as usable as familiar vocabulary embedded in learned linguistic associations.
Read the full abstract ▾
Machine Learning
TRACE: Trajectory Selection for Parallel Scaling of Search Agents
Qisheng Zhou, Zhen Xiong and Qiaoyu Tan
Parallel search may generate a correct answer that final-answer voting fails to select. We formulate this consolidation stage as trajectory selection and introduce TRACE (Trajectory Ranking with Aggregated Cross-Rollout Evidence), a lightweight learned selector that ranks completed trajectories using the search evidence behind their answers. TRACE preserves individual query and evidence occurrences, connects rollouts through shared content or document identity, and propagates information across these relations. Each candidate answer then reads the updated states of its own trajectory, preserving
retrieval provenance while incorporating evidence from related rollouts. Trained with answer-level supervision over frozen text
embeddings, TRACE returns an existing answer without additional search or autoregressive aggregation. One selector per search setting transfers across rollout policies and
agent backbones without
agent-specific fine-tuning, improving over voting across six WebQA policies and six long-horizon
dataset-backbone combinations at $K=16$. On Qwen2.5-14B Base/SFT WebQA pools, TRACE achieves 45.2/49.2% EM, compared with 43.9/48.0% for the strongest Qwen3-32B generative aggregators. On long-horizon FRAMES,
GAIA, and BrowseComp, it reaches 78.6% average accuracy, exceeding majority voting by 3.1 percentage points. On Base WebQA pools, TRACE with only 8 rollouts comes within 0.4 points of majority voting over 64. TRACE also achieves at least $10\times$ higher processing throughput than SolAgg, SummAgg, and AggAgent across all seven WebQA
benchmarks. These results show that reusing cross-rollout search evidence provides an effective and efficient alternative to heavyweight generative aggregation for parallel search. Code is available at https://github.com/Jaasssoooonnnnn/TRACE.
Read the full abstract ▾
Machine Learning
Patient-Centered Treatment Planning for Chronic Multimorbidity: A Hierarchical Reinforcement Learning Framework for Preference Modeling
Nafiseh Payani, Soham Das, G. Anthony Wilson et al. (+1 more)
Patient preference, defined as a patient's demonstrated willingness and capacity to adhere to clinical recommendations, is a primary determinant of therapeutic effect yet remains structurally absent from existing computational treatment planning models. We address this gap by presenting patient-centered factored-action hierarchical option-critic (FAHOC), a hierarchical reinforcement learning (HRL) framework that jointly learns high-level options corresponding to therapeutic strategies and factored intra-option policies that decompose the joint action space into disease- and intervention-specific subcomponents, while imposing a cooperation-aware action masking mechanism. This enables structured exploration, improved credit assignment across hierarchy levels, and more interpretable decision pathways, while enforcing patients' preferences. Formal guarantees establish that cooperative patients achieve higher optimal expected health outcomes than non-cooperative patients, and that the factored Q-function approximation error is provably bounded. The framework is evaluated using longitudinal data collected from approximately 50,000 comorbid hypertension and type 2 diabetes mellitus patients from five hospitals in the Southeast U.S. FAHOC achieves a quality-adjusted life year expectancy equivalent improvement of 0.669 (vs -0.133 observed clinician practice), correctly identifies cooperative patients in 95.9% of cases and never violates a patient's preference in held-out test, demonstrating that HRL with explicit preference constraints can support preference-consistent, clinically safe decision-making in multimorbidity management.
Read the full abstract ▾
All authors Nafiseh Payani, Soham Das, G. Anthony Wilson, Anahita Khojandi
Software Engineering
DoGBench: Can Agents Meet Expert Standards for User-Facing Documentation?
Frances Liu, Manny Silva, Paige Calvert et al. (+2 more)
We introduce DoGBENCH (Documentation Generation
Benchmark), to our knowledge, the first
benchmark for generating and maintaining real user-facing software documentation. It asks whether an
agent can produce documentation that experienced technical writers would accept in review. The
benchmark contains 292 items from open source projects, including Helm, PostHog, and Mautic. Each item gives the
agent a pre-change repository and a trigger, such as a code pull request or a reported documentation gap. The
agent must first decide whether the documentation needs an update. For items that need one, the
agent must produce an acceptable patch in one attempt. For items that do not need updates, the
agent must abstain. Task-specific rubrics, validated with project maintainers, score each patch on accuracy, completeness, reader guidance, placement, and repository conventions. The composite score combines patch quality with correct abstention, and a score of 100 means an
agent meets every requirement for the task. Scores should not be interpreted as a percentage of an expert's capability. We evaluated seven
agents. The highest-scoring
agent reached 47.3 out of 100 on the 117-item held-out split. In a separate audit of 1,267 patches, the most common failure modes were task-completion gaps (45.5%), technical inaccuracies (36.6%), and incomplete conceptual or reference coverage (32.5%). Analysis of the corresponding trajectories identified three key patterns associated with these failures: (1) describing interfaces without examining how readers use them (36.0%), (2) missing decisive evidence and filling the gaps with plausible assumptions (33.1%), and (3) stopping after finding the first plausible documentation surface and leaving other affected pages stale (30.1%).
Read the full abstract ▾
All authors Frances Liu, Manny Silva, Paige Calvert, Ayu Adiati, Sarah Sanders