Artificial Intelligence
OmniVCBench: Benchmarking Evidence-Grounded Multimodal Reasoning Towards AI Virtual Cells
Manyu Li, Xunkai Li, Yongfu Xiong et al. (+3 more)
Artificial Intelligence Virtual
Cells (AIVCs) are envisioned as scientific
agents that simulate cellular responses, explain underlying mechanisms, and support hypothesis-driven discovery. Existing AIVC
benchmarks, however, operate primarily at the simulation layer, motivating complementary evaluation of how models interpret experimental evidence and formulate biological hypotheses. We introduce OmniVCBench, a figure-centric, source-traceable
benchmark for the interpretation component of an AIVC. It contains 6,077 curated single- and multi-subfigure question–answer pairs derived from figures and experimental contexts in the scientific literature. Guided by Bloom's taxonomy, we instantiate interpretation-layer counterparts of the AIVC Predict–Explain–Discover agenda through three scientific
reasoning tasks. We further introduce AIVC-Judge, a task-conditioned MLLM-as-a-judge framework with category-specific, reference-aware rubrics for evaluating open-ended responses. A complementary Model-Derived Hard-Negative Mining (MDHNM) strategy converts plausible errors observed during model inference into MCQ distractors for lower-cost evaluation. Within the evaluated heterogeneous model pool, MCQ accuracy correlates positively with AIVC-Judge scores, providing a complementary view of performance alongside open-response evaluation. Code and data demo are available at https://anonymous.4open.science/r/OmniVCBench.
Read the full abstract ▾
All authors Manyu Li, Xunkai Li, Yongfu Xiong, Yi Liu, Rong-Hua Li, Guoren Wang
Robotics
Urgent Actions Go First: Urgency-Aware Denoising for Real-Time VLA Control
Zibo Wang, Haochen Han, Pengzhen Ren et al. (+2 more)
Diffusion and flow-matching
Vision-
Language-Action (VLA) policies generate action chunks through iterative denoising, incurring substantial inference latency that severely limits real-time robotic control. Existing acceleration methods treat an action chunk as a monolithic computational unit, ignoring a crucial physical reality of receding-horizon control: actions are generated jointly but consumed sequentially, resulting in inherently heterogeneous execution urgencies. We exploit this asymmetry to introduce Urgency-Aware Denoising (UAD), a novel inference-time framework that allocates denoising computation according to when each action is physically needed. UAD releases time-critical urgent actions after fewer denoising steps while overlapping the continued background refinement of tail actions with physical execution. However, heterogeneous denoising introduces two key challenges: early-release errors in urgent actions and trajectory inconsistency in tail actions. UAD elegantly resolves both through two core mechanisms: Trajectory Reconciliation, which reconstructs unified internal state
evolution to restore joint denoising coherence without additional model evaluations, and Ghost Action Correction, which leverages non-executed ghost continuations to dynamically compensate for early-release errors across remaining executable actions. Extensive evaluations across multiple VLA architectures, simulation
benchmarks, and real-world manipulation tasks demonstrate that UAD achieves up to a 1.89x speedup in average action availability latency while maintaining comparable success rates to vanilla inference with optimal denoising budget, offering a more favorable success-latency trade-off than state-of-the-art VLA acceleration baselines.
Read the full abstract ▾
All authors Zibo Wang, Haochen Han, Pengzhen Ren, Mingtong Dai, Fangming Liu
Robotics
Faster and Better? Benchmark Bugs and Design Limitations Distort the Evaluation of Vision-Language-Action Acceleration
Qiwei Chen, Kaijun Zhou, Nuohui Shi et al. (+3 more)
Simulated manipulation
benchmarks are the standard tool for evaluating
vision-
language-action (VLA) policies and the acceleration methods that reduce their inference latency for on-
robot deployment. On these
benchmarks, we observe that some training-free acceleration methods, which approximate the baseline policy's computation, achieve higher measured success rates than the baseline itself. Success rates alone cannot establish whether such gains come from better task execution or from evaluation flaws. We therefore investigate two kinds of
benchmark flaws behind these gains: bugs, where the implementation does not match the intended task or evaluation protocol, and design limitations, where success criteria and simulation settings do not fully capture how acceleration affects task execution. Starting from tasks with anomalous gains, we localize root causes by plotting object trajectories against checker acceptance regions, and classify the resulting bugs into task consistency, initialization, and reproducibility. Extending this audit to seven
benchmarks, including RoboTwin, LIBERO-Plus, and VLABench, we identify 22 bugs of these types and 4 design limitations. For the latter, we revise permissive success checkers, correct unrealistic object masses, and add a motion-aware score that favors smoother actions. Experiments show that bug fixes can reverse method rankings, moving the baseline from last to first on one task. Addressing design limitations can likewise remove anomalous gains: on another task, the baseline moves from 21 percentage points behind an accelerated method to 5 points ahead. Gains attributed to acceleration can therefore be artifacts of the
benchmark rather than better task execution. We release our bug fixes and revised
benchmark settings to support trustworthy evaluation of VLA acceleration.
Read the full abstract ▾
All authors Qiwei Chen, Kaijun Zhou, Nuohui Shi, Zhiyang Li, Yuxuan Feng, Jinyu Gu
Numerical Analysis
Computing Spectral Properties of Differential Operators on Surfaces
Gustav Conradie, Matthew J. Colbrook and Daniel Fortunato
Computing spectra of differential operators on surfaces is crucial in many applications, ranging from medical image analysis to nanoscale
quantum structures. However, discretization of the infinite-dimensional operator and the underlying surface can lead to inaccurate and misleading results. This paper introduces the SurfSpec suite of
algorithms to compute
eigenvalues and spectral measures of operators on surfaces with error control and convergence guarantees. We combine methods based on complex contour integration and residuals to compute discrete spectra and high-frequency eigenmodes, including of nonlinear
eigenvalue problems. We then compute convolutions of spectral measures with rational kernels to deal with continuous spectra. In both cases, the key tool is the operator resolvent applied to functions, which is computed using high-order methods that solve surface PDEs. We illustrate our
algorithms on a variety of differential operators and surfaces, including a triceratops.
Read the full abstract ▾
Cryptography and Security
Selective Channel Restoration for Backdoored Vision-Language Models
Shuming Liu, Zhifang Zhang, Suqin Yuan et al. (+3 more)
Vision-
language models (VLMs) exhibit strong
multimodal capabilities but remain vulnerable to
backdoors implanted through poisoned fine-tuning data. Existing defenses often require extensive parameter updates during fine-tuning or incur per-query overhead during inference. To address these limitations, we propose Perturb-Select-Restore (PSR), a post-training defense that performs sparse updates to the projection interface and introduces no additional computation during inference. We reveal that backdoored VLM projectors are substantially more sensitive to bounded perturbations than clean VLM projectors, a phenomenon we term projection fragility. Building on this finding, PSR identifies the output channels most sensitive to perturbations in each projection layer of a backdoored VLM and restores their parameters to the corresponding pretrained values. Experiments across multiple tasks show that PSR reduces attack success rates to near zero while preserving clean-task performance.
Read the full abstract ▾
All authors Shuming Liu, Zhifang Zhang, Suqin Yuan, Khin Mi Mi Aung, Zhuoyi Lin, Lei Feng
Numerical Analysis · Cross-listed from physics.flu-dyn
Assessment of subgrid scale mixing models used in LES at high pressures
Neelakantan Padmanabhan and Richard S. Miller
Molecular mixing models for the conditional scalar
diffusion (CSD) term in the filtered mass density function (FMDF) transport equation used in Large Eddy Simulation (LES-FMDF) approach are evaluated against direct numerical simulation (DNS) of spatially evolving jets. A database of DNS of turbulent mixing slot jets of $C_{7}H_{16} / O_2$ and $C_{7}H_{16} / Air$ is developed. The formulation includes the compressible form of the governing equations, a generalized multicomponent
diffusion model, a cubic real gas equation of state and realistic property models. Simulations are conducted over a wide range of initial pressures ($1 atm < P_0 < 100 atm$) and jet width based Reynolds numbers of 850 and 1300. FMDF of mixture fraction at various filter widths are obtained from the simulation at several spatial locations within the flow. The CSD term in the exact transport equation of FMDF is calculated from the DNS. An a priori analysis of Interaction by Exchange of Mean (IEM), Modified Curl (MC) and Mapping Closure (MAPPING) mixing models for CSD is conducted. At high pressures, significant deviations are observed between the CSD measured from the DNS and those predicted by the models. The source of the deviations is determined to be the mixing frequency used in the models. The significance of the mixing frequency and its variation with change in pressure and
diffusion models is studied. The primary objective of this work is to assess the applicability of aforementioned mixing models to flows at large pressures and determine the causes for deviation in prediction.
Read the full abstract ▾
Information Theory
Innovation-Based Sampling for New Information
Jiping Luo, Anthony Ephremides and Nikolaos Pappas
This work introduces innovation-based sampling of continuous
stochastic processes, in which a sample is generated only when the process reveals “sufficiently new” information relative to all previous samples. The resulting innovation process admits a
lattice structure and a three-dimensional state representation. For a Wiener process, we characterize the direction, timing, and frequency of innovations. We show that direction reversals become increasingly rare and establish limit
theorems for their number. Innovations also become progressively sparser: the expected number of innovations grows only as the square root of time, and hence the sampling rate vanishes asymptotically. We then study remote estimation from sparsely received innovations. Notably, the absence of an innovation carries information: the minimum mean-square error (MMSE) estimate evolves with the age of information (AoI) and achieves a substantial MSE reduction over the conventional silence-ignorant zero-order hold (ZOH)
estimator. Finally, we develop tractable affine-age and exponential-age approximations for practical use. Overall, information is conveyed not only by the content of innovations, but also by their direction and timing.
Read the full abstract ▾
Computation and Language
Which papyrus HTR is good enough? Character-error-rate tolerance of four papyrological tasks on Greek texts
Anton Repushko and Elena Chepel
Purpose: Most Greek papyri remain unpublished and undigitised; a handwritten text recognition (HTR) pipeline that transcribes them automatically would let scholars discover documents and literary works that have so far gone unread. Recognition systems for Ancient Greek papyri are in statu nascendi, and how accurate they must be for a given papyrological task has not been examined. To answer this and set a
benchmark for Greek papyrus HTR, we test a range of character error rates (CER) against four papyrological tasks, using published editions as ground truth. Methods: From 63,846 current editions of Greek texts in papyri.info, we imitate a letters-only "perfect HTR" output by removing the editorial layer, then degrade it with a seeded
algorithm to exact CERs of 1 - 50%, with lost lines and four error-shape variants. On these data we train small models (TF-IDF, fastText, a character CNN, ByT5-small) for document type, dating and documentary-versus-literary classification, and apply eight keyword search methods. We compare models trained on clean text with models retrained at a specific CER level, and evaluate across CERs. Results: Tolerance differs by task. With clean-trained models, documentary-versus-literary classification retains 90% of its metric up to 20% CER; document type up to 7.5%; subtypes and search up to 5%; dating only up to 3%. Retraining on text containing character errors largely eliminates the sharp degradation that otherwise sets in above 15% CER. Models generally tolerate concentrated damage in a long document better than small errors spread across a short text. Conclusion: The study provides a CER target for each of the four tasks and shows that models trained on noisy text make current, imperfect text recognition useful for them.
Read the full abstract ▾
Artificial Intelligence
Cross-Entropy Guided Routing in Mixture-of-Experts Large Language Models
Yury Nahshan, Nati Daniel, Jacob Goldberger et al. (+1 more)
Sparse mixture-of-experts (MoE) large
language models scale model capacity by routing each
token to a small subset of experts. Their routers are regularized with load balancing terms and learn affinity scores through the
language-model objective. However, these objectives do not provide direct
alignment between routing affinities and
token-level error. We introduce
token-error supervision for sparse routing in two forms. The first form predicts an error score per expert. The affinity-weighted aggregate of these scores is aligned to the next-
token cross-
entropy loss, while the individual scores attenuate affinity before top-$K$ selection. The second directly aligns the router's affinities to the model's objective without requiring an additional head or inference-time modification. Both formulations use the Itakura–Saito divergence or an exponential negative log-likelihood for aligning affinities and
token errors. Across two sparse MoE backbones and four multiple-choice question-answering
benchmarks, we evaluate both supervision mechanisms. On Granite, our method improves accuracy by approximately 2.3 percentage points on average over a parameter-matched routing baseline. With stronger supervision, the gain on ARC-Challenge reaches 2.94 points. Both mechanisms preserve the native sparse execution budget and aggregation policy. Our code is available in the supplementary materials.
Read the full abstract ▾
All authors Yury Nahshan, Nati Daniel, Jacob Goldberger, Yoli Shavit
Computer Vision
Multi-Site Real-World Performance of Commercial AI for Pulmonary and Incidental Pulmonary Embolism Detection
Aawez Mansuri, Mohammadreza Chavoshi, Theodorus Dapamede et al. (+13 more)
Pulmonary embolism (PE) is a leading cause of cardiovascular mortality, yet the real-world performance of FDA-cleared AI detection models remains incompletely characterized. We retrospectively evaluated two FDA-cleared AI
algorithms from a single commercial platform (Aidoc Medical BriefCase), one for PE triage on dedicated CT pulmonary angiography (CTPA; n = 30,678) and one for incidental PE (iPE) detection on routine contrast-enhanced CTs (n = 37,191), across a 17-facility academic health system. Reference-standard labels were extracted from radiology reports using a validated LLM pipeline (97% accuracy, kappa = 0.94). The PE model achieved 86.8% sensitivity and 99.1% specificity, with sensitivity declining from 99.3% for saddle emboli to 72.9% for subsegmental PE, and from 89.7% for acute to 65.3% for non-acute PE. The iPE model achieved 73.5% sensitivity and 99.8% specificity. Both models demonstrated lower sensitivity than FDA-clearance
benchmarks while exceeding cleared specificity, with diminishing performance for peripheral and non-acute emboli mirroring known human reader limitations and underscoring the need for standardized post-
market surveillance of AI-enabled medical devices.
Read the full abstract ▾
All authors Aawez Mansuri, Mohammadreza Chavoshi, Theodorus Dapamede, Wasif Bala, Beatrice Brown-Mulry, Rohan Isaac, Bardia Khosravi, Hanzhou Li, Frank Li, John T. Moon, Chad Robichaux, Dan I.G. Cohen-Addad, Ninad V. Salastekar, Janice Newsome, Judy W. Gichoya, Hari Trivedi
Information Retrieval
Towards Semi-Automatically Comparing Keyword-Based and Semantic Search Accuracy
Mohamed Ben Salha, Fiete Lüer, Maik Betka et al. (+1 more)
The increasing importance of Information
Retrieval (IR) in managing large
datasets has highlighted significant limitations in traditional keyword-based search systems. Context-aware chat-based search methods, such as
Retrieval Augmented Generation (RAG), have recently emerged, but their evaluation compared to keyword-based systems often relies on subjective user feedback. A rigorous, quantitative comparison between these paradigms remains lacking. This work introduces a novel, preliminary framework to quantitatively assess IR accuracy of search systems that produce different output formats, such as lists and messages. It focuses on two key aspects: the ranking accuracy for keyword-based systems and the completeness of retrieved information for semantic chat-based systems. Our approach enables semi-automatic comparisons of semantic and keyword-based methods using interchangeable equivalence classes tailored to domain-specific contexts (e.g., companies or problems). We validate the framework through an industrial case study, demonstrating statistically significant improvements in context-aware search over keyword-based methods, supported by analyses including the Mann-Whitney U-Test. With its adaptable design, the proposed framework provides a strong foundation for objectively assessing keyword-based and semantic chat-based search methods.
Read the full abstract ▾
All authors Mohamed Ben Salha, Fiete Lüer, Maik Betka, Stefan Wagner
Numerical Analysis
Finite element approximation for the Bessel $(p,s)$-Laplacian
Juan Pablo Borthagaray and José Camilo Rueda Niño
We study the finite element approximation of the Dirichlet problem for the Bessel $(p,s)$-
Laplacian, $\operatorname{div}^{s} \left(|\nabla^{s} u|^{p-2}\nabla^{s} u\right)$, built from the Riesz fractional gradient $\nabla^s$ and posed on the Bessel potential space $X^{s,p}$, obtained by complex interpolation between $L^p$ and $W^{1,p}$. For $p = 2$ this operator reduces to the fractional
Laplacian, while for $p \neq 2$ it differs from the fractional $p$-
Laplacian arising from real interpolation. Since a direct Galerkin discretization requires reassembling a dense stiffness
matrix at every nonlinear iteration, we propose an augmented Lagrangian formulation based on the projected fractional gradient $B_h = \mathbf{\Pi}_h \nabla^s$, in which the only dense
matrix is assembled once and the nonlinearity decouples elementwise. We analyze the resulting consistency error, establish a priori convergence rates in $X^{s,p}$, conditional on a discrete inf-sup condition for the augmented Lagrangian scheme, and present numerical experiments that illustrate the convergence of the method and its ability to deal with degenerate and strongly nonlinear problems.
Read the full abstract ▾
Machine Learning
Optimizer-dependent training dynamics converge to the same one-third optimal data scaling
Hyunseok Lee, Mihir Basil, Yizhou Liu et al. (+1 more)
Neural scaling, in which loss falls as a power law with training, is central to large
language models, and one recent proposal is that a $1/3$ exponent emerges from learning peaked distributions. That account describes SGD, but models in practice are trained with adaptive optimizers. Here we separate two exponents the $1/3$ account does not distinguish: how fast the loss falls with training steps along a single run, and how fast the optimally tuned loss falls with
dataset size $D$. We show that the first, a dynamic exponent, is optimizer-specific while the second, an optimal data exponent, converges to $1/3$ across optimizers. In an online teacher-student model we decompose the loss into norm growth (radial) and
alignment toward the teacher direction (tangential), each decaying as a power law with dynamic exponents $\alpha_{r}$ and $\alpha_{t}$. Under SGD, both are close to $1/3$, so the data exponent is also $1/3$ across different learning rates. Under Adam the two separate: $\alpha_{r} \simeq 0.48$ but $\alpha_{t} \simeq 0.08$. Since the total loss is minimized when these two parts are balanced, the optimal learning rate is optimizer-dependent: $D$-independent for SGD but falls with $D$ for Adam. Yet tuned to that optimum, the loss returns to $D^{-1/3}$ for both. A
stochastic-dynamics analysis explains why: the optimizers can trade decay speed between the two channels, but they all fall on a single dynamic exponent relation, $2\alpha_{r}+ \alpha_{t} = 1$, which fixes the optimal data exponent at $1/3$. Across seven optimizers, including
Muon, the measured exponents are consistent with this relation, and the optimal-loss envelopes agree with $D^{-1/3}$ across them. The optimizer sets how fast a model learns per step; tuned optimally, it changes the prefactor but not the rate at which loss falls per sample.
Read the full abstract ▾
All authors Hyunseok Lee, Mihir Basil, Yizhou Liu, Jeff Gore
Artificial Intelligence
ContextRender: From Execution Dependencies to Agent Context
Savini Kashmira, Jayanaka L. Dantanarayana, Lingjia Tang et al. (+1 more)
LLM
agents performing long-horizon tasks accumulate tool results that later steps may need. Passing the full history to every invocation is costly even when it fits within the context window, while reducing it risks omitting needed information. Existing context management methods can overlook how earlier tool results are used in subsequent execution, leaving needed information out of context. We introduce ContextRender, which manages context through a persistent graph of execution dependencies. We develop Tool-Flow Analysis to track how later operations reuse information from earlier tool results, providing a signal called observed reuse. A renderer combines this signal with recency and semantic relevance to select results within a fixed history budget, retaining omitted results for later use. Across AppWorld and 8-objective QA with three execution models, ContextRender outperforms the evaluated context management baselines using a 6K history budget, well below the models' maximum context windows. Within this budget, it achieves task performance close to or above that of passing the full history while reducing mean inference cost by 10.2%-32.2% relative to Full history. Ablations show that observed reuse improves task performance and retention of results reused later.
Read the full abstract ▾
All authors Savini Kashmira, Jayanaka L. Dantanarayana, Lingjia Tang, Jason Mars
Cryptography and Security
Lights, Camera, Attack: Exploiting Temporal HDR Fusion with Pulsed Light
Alkim Domeke, Michael Kuhr, Roman Gilliatt et al. (+5 more)
Modern cameras widely use temporal High Dynamic Range (HDR) to improve visibility by capturing a sequence of exposures with different integration times and fusing them into a single image. This process implicitly assumes that scene illumination remains sufficiently stable during capture. We introduce FLASH (Fusion-Level Attack by Saturating HDR), an external pulsed-light attack that deliberately attacks this assumption by creating cross-exposure inconsistency before downstream perception. FLASH exploits an algorithmic assumption rather than relying on sensor damage or hardware failure, and requires neither physical camera access, access to raw exposure brackets, knowledge of the fusion
algorithm, nor exact phase lock to the camera. Across eight physical camera platforms spanning embedded, surveillance, photography, smartphone, and automotive use cases, and matched optical controls, FLASH causes pipeline-dependent darkening, overexposure, and visibility loss. This includes extreme-darkening rates of 50.0% on an iPhone 16 Pro and 33.7% on a Wyze Battery Cam Pro. On the Wyze camera, FLASH triggers the system-level low-visibility response in 10/10 trials, compared with 0/10 continuous-light and randomized-frequency flashing controls. In a controlled stationary OpenPilot case study, 23.0% of frames exhibit severe darkening in the traffic-cone target region, with target-background CNR decreasing by up to 90.8%. Under FLASH, the OpenPilot interface also fails to display the system-level path state observed in the corresponding control trials. In a controlled night-only HDR reconstruction stress test, a proof-of-concept exposure-rejection defense reduces median output-brightness deviation by 79.16%. These results show that temporal HDR fusion itself requires security-aware validation of exposure evidence.
Read the full abstract ▾
All authors Alkim Domeke, Michael Kuhr, Roman Gilliatt, Mohammad Hamad, Sebastian Steinhorst, Bing Li, Long Cheng, Mert D. Pesé
Machine Learning
HyDI: A hybrid Deep Learning-Inductive Logic Programming ensemble for multi-label classification
Simon Flügel and Till Mossakowski
While attaining remarkable results for many applications, Deep Learning models are notoriously difficult to explain. This work introduces HyDI, a hybrid ensemble architecture for hierarchical multi-label classification. It combines a Deep Learning (DL) model with rule-based classifiers generated by Inductive Logic Programming (ILP). For leaf classes of the label hierarchy, the rule-based classifiers replace the DL model, leading to more transparent classification results. HyDI is applied to the Chemical Entities of Biological Interest (ChEBI) ontology, providing ILP-generated rules for 314 classes. For these classes, HyDI can generate global explanations as well as local explanations that combine visual and text-based descriptions.
Read the full abstract ▾
Cryptography and Security
Where Do LLMs Decide to Break the Rules? Mechanistic Localization of Prompt Injection Compliance
Rui Wen, Jiayang Liu, Zeyu Yang et al. (+2 more)
When a
prompt injection attack succeeds, a Large
Language Model (LLM) abandons its assigned system role to comply with an
adversarial instruction. While prior work has extensively quantified how often this occurs, we ask a more fundamental question: where inside the network does the model actually decide to break the rules? Using layer-by-layer causal activation patching across five models (4B to 32B parameters), we find a clear dissociation: attack information is linearly decodable from the first layer, yet causal leverage over the model's behavior is negligible until a late-layer bottleneck in the final third of the network. Patching this bottleneck reverses compliance in 77–92% of cases. We show that the compliance mechanism occupies a compact linear subspace (rank-8 in 4B and 14B models, scaling to rank-64 at 32B) and is architecturally stable across varying model families. Finally, we validate our mechanistic account by showing that this causal peak layer is also the representationally optimal site for detecting attacks, outperforming early-layer classifiers that degrade under surface-level obfuscation such as leetspeak substitution. This
alignment between causal leverage and detection performance provides converging evidence that the late-layer bottleneck captures decision-relevant computation rather than merely reflecting an artifact of the intervention.
Read the full abstract ▾
All authors Rui Wen, Jiayang Liu, Zeyu Yang, Jun Sakuma, Lu Sun
Robotics
Recompositional Robotics: Cross-Domain, Open-set, and Lifelong Modularity Beyond Morphology
Steven Swanbeck, Jonathan Salfity, Corrie Van Sice et al. (+2 more)
Research in modular robotics has produced capable approaches allowing a
robot's morphology to change online, with recent efforts also developing approaches to decide which morphology to assume and automatically propagate that decision into the
robot's motion planning and control. These approaches are powerful and increase adaptability in the field. However, an alternative objective is not to build
robots whose structures can change, but
robots whose fundamental capabilities can change, where capability is a joint function across several domains, including kino-dynamics, perception, compute, and high-level coordinating behaviors. A
robot designed to be reconfigured across these domains has a greater capacity to alter its capability than one that can be reconfigured in a single domain. We refer to this cross-domain reconfigurability as integration span, and recognize a complementary measure of the resistance to reconfiguration, which we refer to as integration inertia. Current modular
robots have reduced integration inertia in the structural domain while it remains high in the other domains that contribute to integration span. We assert that the systems that can provide the most utility through reconfiguration in practice are those maximizing span and minimizing inertia and call this general problem recompositional robotics: adaptation over a heterogeneous set of modules including hardware, software, compute, and behavior that abstracts each component by the interfaces it requires and provides such that they can be reasoned over holistically. We define the problem, ground it in two deployed systems and active research efforts, and pose open questions about the future of recompositional robotics.
Read the full abstract ▾
All authors Steven Swanbeck, Jonathan Salfity, Corrie Van Sice, Robert Blake Anderson, Mitch Pryor
Machine Learning · Cross-listed from physics.chem-ph
Foundation Neural-Network Quantum States for Molecular Potential Energy Surfaces in Second Quantization
Lizhong Fu, Jianan Wei, Wenguan Wang et al. (+1 more)
Second-quantized neural-network
quantum states have achieved accurate molecular energies, but extending them across molecular geometries requires a shared representation of the geometry-dependent wavefunction coefficients. We introduce geometry-conditioned foundation neural-network
quantum states for molecular electronic structure in second quantization. A single autoregressive model learns a family of ground states from sparse anchor geometries and provides wavefunctions at untrained geometries without further optimization. Orbital
alignment matches orbital identities and transports their phases, establishing an aligned orbital basis across geometries. Frozen energies reach chemical accuracy at every untrained query geometry for N$_2$, CO, and H$_4$. On additional molecular paths, the energy-trained wavefunctions yield dipoles, quadrupoles, and natural occupations without property labels. Across three paired N$_2$ training seeds, orbital
alignment lowers the mean absolute energy error over all untrained query geometries from 34-37 mHa to 0.049-0.085 mHa. At approximately 1 mHa mean absolute error, frozen evaluation reduces the per-geometry cost by $986\times$ relative to independent optimization, yielding an estimated $25.8\times$ end-to-end GPU-cost reduction on a 161-point N$_2$ grid.
Read the full abstract ▾
All authors Lizhong Fu, Jianan Wei, Wenguan Wang, Honghui Shang
Computer Vision
The Camera Inside the Editor: Reading the Implicit Camera of Image Editors with Painted Calibration Patterns
Sebastian Rückerl
Instruction-based image editors insert objects, restyle scenes and render new viewpoints, but it is unknown which camera they assume when they paint into a photograph. Asked to cover the floor with a checkerboard, an editor paints projective structure from which classical vanishing-point geometry reads pitch, roll, focal length, yaw and, on renders, the principal point, without any training. Unlike a calibrator such as GeoCalib, which estimates the camera of an image, this isolates the camera under which the editor paints. On 120 rendered cameras with exact ground truth, Qwen-Image-Edit-2511 paints tile edges that meet their vanishing points within 0.26 degrees, and its implicit camera matches the true one to 0.8 degrees in pitch and 6% in focal length, more accurately than GeoCalib except in roll. Asked to draw the horizon or mark a vanishing point instead, the editor fails, so this knowledge is revealed by painting and not by the explicit tasks we tried. The implicit camera has two priors: roll is pulled towards level (slope 0.71), and telephoto perspective towards a default of about 30 mm, which roughly matches the camera the models paint without any scene. For Qwen, the priors do not grow when blur removes four fifths of the line evidence. They are stronger on real photographs, and on NYUv2 a shorter wording of the task removes the difference for roll. On photographs from a 24–240 mm zoom lens the painted perspective grows with only 0.62 of the lens's slope, while GeoCalib and MoGe-2 saturate at about 52 and 42 mm. FLUX.1 Kontext and LongCat-Image-Edit are pulled much harder. Finally, from a level camera a camera-control LoRA executes pose commands at only 50–70% of their strength, and a board painted into its output agrees with the camera it produced.
Read the full abstract ▾
Machine Learning
Weights Read and Write Features: Scalable Parameter Decomposition Grounded in Activation Space
Tue M. Cao, Lisiane Pruinelli and My T. Thai
Activation space and parameter space provide complementary views of model computation. Activations represent information, while weights read, transform, and write that information. Yet existing interpretability methods largely study the two spaces separately, leaving the connection between represented information and parameter-level computation underexplored. We introduce Activation-Supported Parameter Decomposition (ASPD), which jointly decomposes activation and parameter spaces and grounds each learned weight component in the activation features it reads or writes. This grounding constrains otherwise non-unique parameter decompositions using the model's internal activations, while an internal reconstruction objective provides a local learning signal at the weight
matrix being analyzed. Together, these properties enable scalable, interpretable, and causally editable parameter decomposition in pretrained large
language models, demonstrated on Qwen-3-8B. The learned read–write components can also be composed into parameter-level mechanism
circuits. We use ASPD to recover mechanisms underlying the classic IOI
circuit and trace semantic transformations through model weights.
Read the full abstract ▾
Artificial Intelligence
Spatiotemporal Hyperedges for EEG Seizure Detection and Prediction
Hyunju Kim, Sheo Yon Jhin, Noseong Park et al. (+1 more)
Seizure detection and prediction from
EEG are clinically important but challenging because seizures are rare, temporally localized, and propagate as coordinated events across multiple channels. Recent dynamic graph neural networks model this by running a temporal model over a sequence of per-time-step pairwise channel edges. However, this pairwise construction misses the spatiotemporal coupling that constitutes a seizure, at substantial training cost. We propose HyBrain, which summarizes spatiotemporal
EEG evidence through a small set of soft hyperedges rather than pairwise edges. A per-channel Mamba backbone produces one
token per (channel, second), and a spatiotemporal hyperedge block pools these
tokens into E_h shared group
embeddings through soft memberships and broadcasts them back. The same encoder serves three downstream tasks: window-based detection, one-second point-wise detection, and preictal seizure prediction. On TUSZ and CHB-MIT, HyBrain achieves the best AUROC on every reported setting against ten baselines, with the largest gap on long-clip preictal prediction. It also matches the most efficient baselines in training time and peak GPU
memory. A qualitative analysis shows that even a single learned hyperedge cleanly captures the preictal -> ictal -> postictal trajectory on a real seizure clip.
Read the full abstract ▾
All authors Hyunju Kim, Sheo Yon Jhin, Noseong Park, Nabil Imam
Logic in Computer Science
Formal Reasoning about Performance Models
Moussa Labbadi, Rupak Majumdar, V.R. Sathiyanarayana et al. (+1 more)
Discrete-event simulation is a standard technique for modelling and analysing the performance of computer systems, networks, and services. Although simulation tools are widely used,
reasoning about the correctness and performance guarantees of the models they implement remains largely ad hoc: simulation outputs are interpreted statistically, but there is no logical foundation for deductive
reasoning about their behaviour. We present a core imperative calculus that captures the essential constructs common to discrete-event simulators: asynchronous execution, continuous and discrete sampling from distributions, and time-based event scheduling through a global event queue. On top of this calculus, we develop a proof system for
reasoning about almost-sure reachability and expected reaching time properties. Our main result is a sound and complete proof rule for these properties. Our framework generalizes deductive
reasoning for discrete-time probabilistic programs to the setting of performance models, in which continuous time and continuous probability distributions are central. We have implemented the proof rules in a tool embedded in Lean. We demonstrate the applicability of our proof rule by deriving proofs of almost-sure reachability and expected reaching time for a number of case studies, including client-server examples that go beyond analytic solutions from queueing theory as well as convergence behaviours in network routing protocols. Establishing the soundness and completeness of our proof rules requires significantly more complex arguments than in the discrete-time setting. This is due to the fundamentally discontinuous nature of the operational semantics and the measure-theoretic challenges of continuous time and probability distributions.
Read the full abstract ▾
All authors Moussa Labbadi, Rupak Majumdar, V.R. Sathiyanarayana, Sadegh Soudjani
Logic in Computer Science
Modelling Shared-Space Coordination in mCRL2: a Bach-to-mCRL2 Translation Framework
Corentin Reuther and Jean-Marie Jacquet
Although significant research has focused on the theory and implementation of data-based coordination
languages, the critical aspect of their automated verification using model-checking techniques remains underexplored, which is essential for ensuring reliability and correctness in distributed systems. Existing tools, such as Anemone, provide a solid foundation for reachability-based verification of Bach programs. While they effectively analyze properties expressed in terms of state attainability, extending support to more expressive temporal specifications-such as liveness properties or invariants over shared space contents-remains an open opportunity. Such extensions are important to capture comprehensive system behaviors, for instance, ensuring that "a request is always matched by a response" or that "no message is silently lost". Addressing this limitation, we propose an automated translation from Bach to mCRL2, which explicitly represents the shared space, thereby enabling the use of mCRL2's mu-calculus model checker to verify complex properties beyond simple reachability. Complementing this translation, we introduce a systematic method to analyze the shared space in mCRL2 concerning data reachability, by directly expressing properties over the shared space's contents in the mu-calculus, thus providing a clearer framework for verification. This approach enables a novel verification process that combines action-based properties with state-based properties over the shared space contents, a largely unexplored area in current coordination-
language verification approaches, offering a new dimension of analysis
Read the full abstract ▾
Artificial Intelligence
Rulin Shao, Shannon Zejiang Shen, Junjie Oscar Yin et al. (+10 more)
We introduce Context
Language Models (CLMs),
language models that natively manage their own context. We implement this by treating the context as a file and allowing the model to make unrestricted updates to this file. This allows the model to learn what is most important to maintain in context, and naturally extends to multi-
agent systems where multiple
agent contexts coexist as files. Building CLMs zero-shot with existing models outperforms SOTA context management strategies across a variety of tasks: 11.4% higher accuracy with 21.5% fewer FLOPs on BrowseComp-Plus, 5% higher scores with 59% fewer FLOPs on 12-hour EdgeBench, and 65% greater improvement with the same compute on a 24-hour multi-repository
agent-swarm task. Moreover, by shifting context management from external harness control to intrinsic model behavior, CLMs naturally enable both in-context and parametric learning of context-management strategies. We show that CLMs can be steered with natural-
language instructions evolved through a standard skill-optimization loop, improving held-out accuracy by up to 35.9 points on a context-management task while reducing compute. We also introduce an online reinforcement learning method for CLMs, improving Qwen3.5-9B performance on BrowseComp-Plus by 47.6% while using 12% fewer FLOPs. Finally, we co-design Suffix Cache Reuse for CLM serving, further reducing server-side compute by 35% relative to standard SGLang at matched performance.
Read the full abstract ▾
All authors Rulin Shao, Shannon Zejiang Shen, Junjie Oscar Yin, Yuetai Li, Minheng Wang, Hamish Ivison, Radha Poovendran, Nathan Lambert, Teng Xiao, Mike Lewis, Wen-tau Yih, Luke Zettlemoyer, Pang Wei Koh
Robotics
CogWAM: Aligning Semantic Cognition with World Action Modeling via Event-Driven Interfaces
Sen Wang, Liu Liu, Xinjiang Wang et al. (+7 more)
Robot policies increasingly incorporate semantic
reasoning and future-world prediction, yet combining these capabilities does not guarantee that local predictions and actions remain aligned with task progress. We introduce CogWAM, a cognition-guided world-action model that establishes an explicit semantic interface between task
reasoning and world-action learning through a persistent Semantic State, which stores completed task events and the active subtask. CogWAM updates this state only when observations indicate semantic transitions, allowing task-level context to persist across multiple action chunks. To bridge semantic context with physical prediction and control, CogWAM employs progress-conditioned WORLD and ACTION queries that selectively extract task-relevant information for future-world prediction and action generation. During training, the Semantic State provides shared task-progress context for both branches, while inference removes the future-prediction branch and directly generates actions from observations and the maintained state. We further introduce semantic training strategies to improve transition learning and closed-loop conditioning. Without additional
robot-action pretraining, CogWAM achieves 15.56 / 11.70 % Score/SR on RoboDojo and state-of-the-art performance on BiCoord, while real-world experiments demonstrate closed-loop dual-arm manipulation with 16.4 fewer Semantic State regenerations than step-wise updating.
Read the full abstract ▾
All authors Sen Wang, Liu Liu, Xinjiang Wang, Zequn Chen, Haoyi Jiang, Taojun Ding, Tingyang Xiao, Zhizhong Su, Jie Wang, Sanping Zhou
Machine Learning
Predictive Geometry of Hidden Trajectories in Transformers
Timur Mudarisov, Mikhail Burtsev, Tatiana Petrova et al. (+1 more)
Decoder-only
transformers are trained only through a terminal next-
token prediction loss, yet this loss constrains every intermediate hidden state through the fixed downstream computation. We formalize this constraint by studying layerwise loss-to-go functions: the terminal loss obtained by continuing a candidate hidden state through the remaining
transformer blocks. Around successful validation trajectories, we show that the local second-order geometry of these functions is governed, up to low-loss residual terms, by a pullback Fisher operator on hidden-state space. Its spectrum identifies output-sensitive directions and approximately prediction-null directions, yielding a local observable subspace of the residual stream. For causal
transformers, the same geometry induces a tokenwise
curvature score: a Fisher-weighted sensitivity of the target logits to perturbations of each
token's hidden state. This score vanishes outside the causal ancestor set of the target and is controlled by downstream Jacobian couplings, making it a loss-aware alternative to
attention magnitude. We estimate these quantities using
matrix-free Jacobian-vector and vector-Jacobian products and evaluate them across decoder-only
language models on WikiText, OpenWebText, and FineWeb. Empirically, the induced geometry predicts perturbation sensitivity, supports nonuniform layerwise rank allocation, yields competitive structured
token-
pruning signals, and improves low-rank student recovery when added to stronger autoregressive
distillation objectives such as reverse KL and skew KL. These results support a predictive-
geometric view of
transformer computation: near successful trajectories, the terminal loss induces a thin, anisotropic set of output-relevant hidden-state directions that can be measured and exploited for compression and
distillation.
Read the full abstract ▾
All authors Timur Mudarisov, Mikhail Burtsev, Tatiana Petrova, Radu State
Machine Learning
Volatility-Clustering Adaptation for Financial Time Series
Manh Nguyen, Minh Hoang Nguyen, Huu Hiep Nguyen et al. (+2 more)
Time-series foundation models are increasingly adapted to new domains through fine-tuning on target data, under the implicit assumption that more target data yields better forecasts. We show that this assumption can fail in financial
forecasting, where individual price changes are difficult to predict, but large moves tend to cluster, creating alternating calm and turbulent periods. Using financial foundation models trained on price bars of open, high, low, close, and volume, we argue that adapting to financial domains requires training signals beyond next-
token prediction. We introduce
Volatility-Clustering Adaptation (VCA), which augments next-
token cross-
entropy with a differentiable penalty on the autocorrelation of squared returns, the standard statistical signature of
volatility clustering. This additional objective provides a multi-step training signal by matching the resulting dependence structure of autoregressive rollouts to those of the realized future. Across three asset sets and two evaluation conventions, VCA improves adaptation over the pre-trained model, with the strongest gains under the primary evaluation (fore), driven primarily by reduced variance error. Overall, our results suggest that effective financial adaptation requires objectives that capture domain-specific temporal structure beyond
token-level prediction.
Read the full abstract ▾
All authors Manh Nguyen, Minh Hoang Nguyen, Huu Hiep Nguyen, Van Dai Do, Hung Le
Computation and Language
Billiger.de Products: A Bilingual Entity Matching Benchmark
Aaron Steiner, Ksenia Elagin, Ralph Peeters et al. (+2 more)
Existing product matching
benchmarks primarily contain English-
language product data and are often dominated by a single product category, such as electronics. This paper introduces Billiger.de Products, a bilingual German and English entity matching
benchmark covering thirteen consumer product categories, including difficult-to-handle categories such as clothing and furniture. The
benchmark data originates from the German price comparison platform billiger.de. Following the design of WDC Products, the
benchmark offers multiple variants that differ in the fraction of corner cases, the size of the development set, and the fraction of entities unseen during training. An aligned English translation of every offer keeps all pairs, splits, and labels fixed, while cross-
language test sets combine German and English records within individual pairs. We validate the
benchmark using six supervised matchers and zero-shot GPT-5.2 on both
language versions and the cross-
language test sets. The validation shows the difficulty of the
benchmark. The comparison of the results on the English version of the
benchmark to the results on the German version shows that most matchers score on average higher on the English version. The difference is largest for RoBERTa and HierGAT, while the zero-shot LLM runs are largely insensitive to the
language. Comparing the F1 scores achieved by PLM-based matchers on the English version of Billiger.de Products with their performance on existing English-
language benchmarks, such as WDC Products and Abt-Buy, shows that Billiger.de Products is more difficult than these
benchmarks.
Read the full abstract ▾
All authors Aaron Steiner, Ksenia Elagin, Ralph Peeters, Johannes Knopp, Christian Bizer
Computer Vision
PolyOCR-Venus: Unified OCR Foundation Models for Text-Centric Visual Intelligence
GuangJian Team: Kaili Huang, Yongshuo Zhang, Bingtao Fu et al. (+21 more)
Optical Character Recognition (OCR) is evolving from plain-text transcription toward general visual intelligence, requiring models to recognize, localize, and reason over textual information in complex visual environments. However, existing OCR systems often excel at only some tasks and struggle to balance recognition, parsing, and
reasoning across scenarios. In this report, we present PolyOCR, a family of unified OCR foundation models of varying scales. PolyOCR combines a shared instruction-following framework with a large-scale data engine that converts heterogeneous visual resources into quality-verified OCR supervision. We introduce Competence-Guided Policy Optimization, which combines verifier-based Group Relative Policy Optimization with on-policy
distillation through sample-wise routing based on teacher reliability and the teacher–student competence gap. We also introduce OCRBench v2.1, our revision of OCRBench v2 with manually verified annotation corrections and task-aligned scoring metrics. Extensive experiments across OCRBench v2.1, CC-OCR, in-house KIE
Benchmark, OmniDocBench v1.6 and MDPBench demonstrate that PolyOCR achieves state-of-the-art or highly competitive performance.
Read the full abstract ▾
All authors GuangJian Team: Kaili Huang, Yongshuo Zhang, Bingtao Fu, Changjiang Jiang, Chenfan Qu, Chenfeng Zhang, Fangming Cui, Gaoyang Zhang, Jiangwei Xie, Jianshu Li, Jing Huang, Jingwen Bai, Mingqi Fang, Tao Fang, Weihong Zhang, Wenbo Du, Xiongfei Bai, Xuekang Zhu, Yinan Xia, Zhenming Wang, Jian Liu, Jingjing Liu, Xiang Qi, Weiqiang Wang