Peer Revenue Past citations are no guarantee of future returns

Papers

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

Computer Science

cs Watch Computer Science

Saturday 19 September

Databases

Same Pattern, Different Answer: A Reference Semantics and Divergence Map for GQL and SQL/PGQ Path Patterns

Madhulatha Mandarapu and Sandeep Kunkunuru

GQL (ISO/IEC 39075:2024) is the first international standard for a graph query language, and SQL/PGQ (ISO/IEC 9075-16:2023) embeds the same pattern-matching core, GPML, in SQL. Both fix a precise semantics for path patterns: four path modes, four selectors, quantified segments. What engines compute for those patterns has never been measured. Work on GQL is theoretical and runs no engine; cross-engine work measures performance and normalises the semantics away. We build an executable reference semantics for the GPML path core and gate it against the six worked answers printed in the standard's own reference exposition. We then run a 17-construct suite against six releases of five products - Kuzu, DuckPGQ, Neo4j at 5.26 and 2026.04, Memgraph, Apache AGE - scoring each cell as conforming, diverging, rejected, or inexpressible in that dialect. Nine of seventeen constructs draw more than one answer across the engines that accepted them. Of 26 disagreements, 15 are silent - the query runs, returns a different multiset, raises no error - a silence ratio of 0.58. Scoring each divergence against the path mode each engine's manual declares separates language from implementation: 9 of 15 are documented differences, 6 are departures from the standard the engine implements. A metamorphic layer needing no reference semantics finds 39 self-consistency violations, yielding two minimal defects. No engine we measured parses the order the standard's own examples are written in, except the authors' own, which added it after this suite reported the gap. Suite, semantics, reproducers and results are open source.
Read the full abstract

Computer Vision

BrainIAC: Interactive 3D Brain Lesion Segmentation across Heterogeneous MRI Modalities with Online Adaptation

Wentian Xu, Anthony P Addison, Ziyun Liang et al. (+3 more)

Brain lesion segmentation is a fundamental task in medical image analysis, playing a critical role in diagnosis, treatment planning, and longitudinal disease monitoring. Yet existing models still struggle to meet the demands of real clinical use, where deployments contain data distribution shifts, arising from differences in scanner hardware, imaging protocol (varying MRI modality sets), and new pathologies. We present BrainIAC (Brain lesion Interactive Adaptive Continuously learning segmentation), a unified framework that integrates (i) a multi-modal backbone network trained to segment multiple types of brain lesions and handle heterogeneous sets of modalities via zero-filling and random modality dropping; (ii) 3D interactive segmentation with bounding-box and click prompts that preserves fully automatic prediction when no prompt is given; and (iii) an online adaptation mechanism combining Mid-Interaction adaptation and Post-Interaction adaptation, supervised by the network's own predictions as pseudo labels and guided by an extra Click-Centered Gaussian loss. To our knowledge, this is one of the first 3D online adaptation methods for interactive segmentation, and the first to combine handling of heterogeneous modality sets with online adaptation. Experiments across seven brain MRI datasets demonstrate that the proposed components provide complementary and synergistic benefits. The method consistently outperforms existing approaches and generalizes well across heterogeneous imaging modalities, including those unseen during training, as well as previously unseen brain pathology types. The code and a 3D Slicer plug-in will be released at https://github.com/WenTXuL/BrainIAC upon publication.
Read the full abstract

All authors Wentian Xu, Anthony P Addison, Ziyun Liang, Harry Anthony, Guang Yang, Konstantinos Kamnitsas

Artificial Intelligence

PINNForge: Execution-Grounded Evolutionary Design of Physics-Informed Neural Networks

Mingyang Yu, Xu Yang, Jun Zhang et al. (+2 more)

Physics-informed neural networks (PINNs) require coordinated choices across representation, architecture, sampling, constraints, loss construction, and optimization, yet effective configurations can vary substantially across partial differential equations (PDEs). Existing automated design methods search over these choices, but training outcomes are still used primarily to rank candidates rather than to inform subsequent designs. We develop PINNForge, an execution-grounded large language model (LLM)-driven evolutionary framework that treats observed PINN training behavior as a cross-generation design signal. PINNForge initializes diverse configurations from PDE-related prior knowledge, executes them, retains globally strong configurations, and feeds their execution evidence and accumulated run-level experience back to the LLM. The LLM then revises and recombines coupled design components or explores new combinations within a generate–execute–evaluate–evolve loop. Across 25 PDE benchmarks, PINNForge achieves the lowest mean MSE on 24 tasks among RoPINN, PINNsFormer, PINNsAgent, and PINNForge. Removing knowledge guidance, execution feedback, or evolutionary search increases the median task-wise MSE ratio to 3.74$\times$, 12.10$\times$, and 10.10$\times$, respectively, relative to full PINNForge.
Read the full abstract

All authors Mingyang Yu, Xu Yang, Jun Zhang, Jing Xu, Keqian Li

Computational Engineering

Shading-Aware Rooftop PV Placement

Tobias von Arx and Srinivasan Keshav

Rooftop photovoltaic (PV) design requires careful panel placement to make effective use of potentially limited roof space and available sunlight. Existing simplified roof models can miss objects and superstructures that shade panels. We simulate rooftop irradiance using hourly weather and a terrain horizon, computing shadows from nearby buildings, trees, and roof superstructures rendered from the Sun's viewpoint. Our method (i) infers existing panel layouts from imagery and estimates their yield, (ii) relocates these panels to improve the yield, and (iii) generates shading-aware layouts for new installations. We compare modeled yield with production records from 11 Swiss installations and evaluate new layouts in simulation on 19 additional rural roofs. Without knowledge of any electrical or system specifications, we obtain a median daily correlation with measured production of 0.982. Absolute energy estimates remain sensitive to system specifications. Comparing new layouts with equal panel counts, accounting for neighborhood and roof-detail shading changes the position or orientation of an average of 38.9% of panels per roof. These layouts increase modeled annual yield by up to 7.91% compared with layouts designed without accounting for neighborhood and roof-detail shadows. We publicly release our tool for modeling and designing rooftop PV installations with terrain, neighborhood, and roof-detail shading at https://github.com/tobiasvonarx/shading-aware-pv.
Read the full abstract

Computer Vision · Cross-listed from eess.IV

SomaNet: Weakly Supervised Learning for Instance Soma Segmentation in 3D Electron Microscopy with Partial Annotations

Mohammad Khateri, Morteza Ghahremani, Jussi Tohka et al. (+1 more)

Soma instance segmentation, i.e., identifying and delineating individual cell somas as distinct instances, is crucial for cellular analysis and connectomic reconstruction. Three-dimensional electron microscopy (3D EM) provides nanometer-scale resolution for capturing fine-grained soma morphology. However, dense instance-level manual annotation is prohibitively costly, limiting the scalability of fully supervised methods. To address this challenge, we propose SomaNet, a weakly supervised framework for 3D EM soma instance segmentation under partial annotation constraints. SomaNet adopts a teacher–student learning paradigm tailored to partial labels. The teacher is trained using partially annotated data to generate pseudo-labels, while the student jointly learns from the partial ground-truth annotations and the generated pseudo-labels, progressively recovering dense instance segmentations. To accurately delineate cell somas under limited supervision with varying instance counts, SomaNet incorporates affinity learning, which encourages high similarity within instances and low similarity across instance boundaries. Semantic-guided affinity decoding and 2D-to-3D reconstruction then produce volumetrically consistent 3D soma instances while preserving the large receptive fields of 2D backbones. The framework is architecture-flexible and supports diverse backbones, including vision transformers and foundation models, enabling direct transfer of pretrained visual representations to volumetric EM segmentation. Experiments on 3D EM brain datasets demonstrate that SomaNet achieves accurate and robust soma instance segmentation across regions with diverse soma morphologies under partial annotation. Code is available at https://github.com/mkhateri/SomaNet.
Read the full abstract

All authors Mohammad Khateri, Morteza Ghahremani, Jussi Tohka, Alejandra Sierra

Computer Vision

Reconstructed holograms and explanation-aware evaluation for low-cost computational pollen analysis in veterinary cytology

Swarn Warshaneyan, Joial Danyal, Blaž Cugmas et al. (+4 more)

Automated pollen analysis supports veterinary cytology, but brightfield microscopy is costlier and more complex than lens-less digital in-line holographic microscopy. We evaluate whether reconstructed holograms can narrow this gap and whether model explanations remain reliable under modality change. Six pollen species were imaged by brightfield and holographic microscopy. Raw, single back-propagation and iterative phase retrieval holograms were evaluated with YOLOv26s detection and MobileNetV4 classification after anchor-based annotation transfer. Six attribution methods were assessed for spatial grounding and faithfulness with the Attribution Health Inspection and Repair (AHIR) protocol, which tests model brittleness under weak noise and corrects attribution-map granularity when needed. Brightfield achieved 0.6890 mAP50-95 (0.8865 mAP50) for detection and 0.9687 macro-F1 (0.9705 accuracy) for classification. Reconstructed holograms narrowed the gap with a task-dependent split: p-type was strongest for detection at 0.5324 mAP50-95 (0.8229 mAP50), while r-type was strongest for classification at 0.7695 macro-F1 (0.7866 accuracy), both far above raw-hologram baselines. Activation-based explanations localized strongly on grains, and region-based methods retained ~60 to ~80% of faithfulness under holography. The holographic detector was highly brittle to weak perturbations, saturating deletion-based evaluation while insertion remained informative. Pixel-level gradient explanations approached random floor, yet spatial smoothing restored p-type gradient faithfulness from 0.05 to 0.51. For holographic classification, perturbation-based explanations remained faithful while gradient-based methods fell below random floor. Reconstruction improves low-cost holographic pollen analysis, while AHIR distinguishes genuine attribution failure from artifacts caused by model brittleness and map granularity.
Read the full abstract

All authors Swarn Warshaneyan, Joial Danyal, Blaž Cugmas, Mindaugas Tamošiūnas, Edgars Kviesis-Kipge, Kirishanth Manivannan, Roberts Kadiķis

Artificial Intelligence · Cross-listed from quant-ph

Watching Quantum Models Think: Hilbert-Space Interpretability in Quantum Transformer Blocks

Diego Iacopetta and Andrea Gasparini

Deep learning models are powerful but opaque. As quantum machine learning matures, the field faces a defining choice: build quantum models that are equally opaque, or exploit the mathematical structure of quantum mechanics to make them inherently interpretable. We show that the latter is possible. By tracking quantum mutual information~(MI), entanglement entropy, and state fidelity through the layers of a Quantum Transformer Block (\qtb{}), a fully-coherent variational circuit with quantum analogues of both attention and feedforward, we gain direct insight into how the model processes information: which tokens it attends to, when correlations form, and why predictions fail. On four tasks with known dependency structure we show that (i)~learned MI matrices align with ground-truth task structure (AUC$\,{=}\,0.69$ on lookup), (ii)~disabling entangling gates collapses accuracy from 100% to 15% while MI$\to 0$, proving entanglement is the mechanism, (iii)~accuracy and MI co-evolve during training ($\rho\,{=}\,0.92$ on lookup), and (iv)~per-sample MI predicts prediction correctness on the conditional task with ROC AUC$\,{=}\,0.84$. All results are validated on IBM Quantum hardware (ibm\_kingston, Heron~r2): the circuit's reasoning process, from product state through structured entanglement, is directly observable on a superconducting processor. These proof-of-concept results, obtained on small synthetic tasks, suggest that the physics of quantum computation can provide intrinsic interpretability signals with no direct classical counterpart, motivating study of whether this advantage persists at scale.
Read the full abstract

Computational Complexity

An exponential lower bound for the bit pigeonhole principle in resolution over parities

Kamil Braun

Resolution over parities, $\mathrm{Res}(\oplus)$, is the characteristic-two version of resolution over linear equations: clauses are disjunctions of affine equations over $\mathbb F_2$. Superpolynomial size lower bounds were previously known only for restricted refutations: tree-like, regular, or of bounded depth. We prove that every DAG-like $\mathrm{Res} (\oplus)$ refutation of the bit pigeonhole principle with $n+1$ pigeons and $n=2^\ell$ holes has more than $\exp(n/(32768\ell^2))=2^{\Omega(n/\log^2 n)}$ clauses, for every $\ell\ge32$, with no restriction on regularity or depth. The proof translates an arbitrary refutation with $S$ clauses into a polynomial calculus refutation of degree $O(\log n)$ over $O(S+n^2)$ groups of extension variables in the style of Buss, Impagliazzo, Krajicek, Pudlak, Razborov, and Sgall. One substitution then removes all extension variables at once and leaves a nonzero low-degree polynomial derived from the pigeonhole axioms alone at degree at most $n/2$; a degree lower bound in the style of Razborov, proved through the homology of chessboard complexes, shows that no such derivation exists. The argument also yields a general sufficient condition for $\mathrm{Res}(\oplus)$ size lower bounds. The main theorem, this condition, and all their dependencies are formalized in Lean 4, and every statement links to its formal proof. The proof was developed with substantial AI assistance within an open research framework described in the final section.
Read the full abstract

2609.23015 · 19 Sep · Main results and their dependencies formalized in Lean 4; verification map and reproduction instructions included. 43 pages. v2: exposition revised with a running example proof DAG, worked algebraic examples and an illustrated matching-moment filling; theorem statements and Lean proofs unchanged

Abstract PDF $PLYN ×2$THRM ×1

Computational Engineering

Cognitive behavior emerging in mechanical systems through reservoir encoding and active inference

Matteo Torzoni, Domenico Maisto, Andrea Manzoni et al. (+3 more)

Mechanical systems are traditionally designed as passive assets, where sensing, analysis, and control are treated as decoupled and externally prescribed processes. Reliance on predefined mechanistic models and supervised learning with offline optimization further limits their adaptability in changing environments. Drawing inspiration from biological cognition, we introduce a computational framework enabling mechanical systems to learn, adapt, and make decisions through interactions with partially observable environments. Perception, action, and learning are integrated through active inference within a perception-action loop. Sensor data are assimilated through a neural reservoir to support continual belief updating, while actions actively probe and influence the physical system, becoming an integral part of the inference process. Through self-directed interaction, adaptive behavior emerges from the continual refinement of internal representations to explain varying observation patterns while pursuing mechanical objectives. Across simulated case studies involving multiple mechanical systems, static and dynamic sensing, and objectives ranging from deformation maximization to vibration mitigation, we demonstrate the potential of the proposed framework as enabling paradigm for autonomous mechanical systems.
Read the full abstract

All authors Matteo Torzoni, Domenico Maisto, Andrea Manzoni, Francesco Donnarumma, Giovanni Pezzulo, Alberto Corigliano

Computer Vision

CrowdCue: Specialist-Cue Conditioning for Vision-Language Crowd Counting

Moshiur Farazi, Bekir Ciftler, Abdulhalim Dandoush et al. (+1 more)

Generative vision-language models (VLMs) offer a counting paradigm in which one model produces both a count and a natural-language account of the scene, yet their raw counting accuracy sits in the range of sub-million-parameter specialist regressors. The open question is whether auxiliary guidance from a pretrained specialist can lift them into useful territory, and through which channel that guidance is best routed. We evaluate Qwen2.5-VL-7B on four widely used crowd counting benchmarks (ShanghaiTech A and B, UCF-QNRF, NWPU-Crowd). Zero-shot prompting rarely produces a parseable count, so LoRA supervised fine-tuning establishes the baseline at overall MAE 81.64. Conditioning on a P2PNet-derived density heatmap as an auxiliary visual signal fails in every encoding we tested, and an adversarial-swap protocol shows the model reads the heatmap but applies it counterproductively. We propose CrowdCue, a family that supplies the same specialist's already-integrated integer count to the VLM as a discrete symbol. The text-channel variant reaches MAE 72.04. The visual-channel variant, which renders the integer as printed digits and supplies it as a second image, reaches MAE 62.65, the strongest result in this paper and well ahead of the cue-supplying specialist alone (84.45 on the same split). In the late-fusion VLM we study, the binding constraint is not the channel but the abstraction level at which the specialist signal is delivered.
Read the full abstract

All authors Moshiur Farazi, Bekir Ciftler, Abdulhalim Dandoush, Reda Bendraou

Information Theory

Hamiltonian Monte Carlo for Vector Perturbation Precoding in MU-MIMO via Continuous Relaxation

Junichiro Hagiwara, Toshihiko Nishimura, Yasutaka Ogawa et al. (+1 more)

Multi-user multiple-input multiple-output (MU-MIMO) is a key technology that improves wireless capacity through multiple antennas. In MU-MIMO downlink precoding, vector perturbation (VP) is a representative nonlinear method that achieves high performance. However, its search for the integer perturbation vector reduces to a closest vector problem, whose complexity grows rapidly as the number of users increases. We propose a method that relaxes the discrete structure of the integer perturbation into a continuous mixture of $t$-distributions, enabling efficient search via gradient-based Hamiltonian Monte Carlo (HMC). Complexity analysis and numerical experiments demonstrate the effectiveness of the proposed method. Its search complexity scales as O(N^2) in the number of users N. At a symbol error rate of 10^-3, it performs within 2.4 dB of a hypersphere approximation benchmark, which approximates the performance limit of VP. This paper reframes the VP perturbation search as a probabilistic inference problem, providing a general formulation for handling high-dimensional discrete search in a continuous space.
Read the full abstract

All authors Junichiro Hagiwara, Toshihiko Nishimura, Yasutaka Ogawa, Takeo Ohgane

Computer Vision

MixiMotion: One-Step Text-to-Motion Generation via Asymmetric Set Distillation

Hung Dinh, Binh Mai, Tran Quoc Bao Le et al. (+2 more)

Iterative text-to-motion generation delivers high-quality and semantically aligned motions but requires multiple network evaluations, resulting in substantial inference latency. We present MixiMotion, a strict one-step text-to-motion generation framework based on offline set distillation. Instead of distilling a single teacher trajectory for each text prompt, MixiMotion constructs an offline bank of multiple teacher motions and aligns teacher and student sample sets through asymmetric bidirectional matching. The teacher-to-student direction promotes coverage of diverse teacher-supported motions, while the student-to-teacher direction suppresses unsupported generations. We further introduce differentiable decoded-space kinematic supervision to complement normalized representation matching with constraints in the decoded motion space. At inference, MixiMotion generates a complete motion sequence with a single network evaluation, without teacher queries, iterative sampling, or candidate ranking. On ViMoGen, MixiMotion achieves a semantic alignment score of $0.835$, outperforming the evaluated one-step baselines and approaching the $0.858$ score of its 50-step HY-Motion-1.0-Lite teacher. In blinded human evaluation, MixiMotion obtains an overall rating of $4.33$, compared with $4.50$ for the teacher, while outperforming the evaluated one-/few-step baselines. Meanwhile, generation latency is reduced from $829.58$\,ms to $9.30$\,ms, corresponding to an $89.2\times$ speedup. These results demonstrate an effective quality–efficiency trade-off for strict one-step text-to-motion generation.
Read the full abstract

All authors Hung Dinh, Binh Mai, Tran Quoc Bao Le, Lam Nguyen, Cong Tran

Programming Languages

A Linked List of Cases in Language Design

Bo Joel Svensson and Benjamin Vedder

LispBM is a Lisp dialect for microcontrollers and embedded systems. This paper is about the changes made to LispBM resulting from it being integrated into the VESC family of firmware and thus gaining a userbase. As a surprise to Author 1, the new users of LispBM wrote programs larger than ever imagined. This used up all available heap space and we were forced to find a way to evaluate Lisp programs partitioned across RAM and flash memory. This story and a couple more, like the one about the Lisp-image system for fast boots, are the core contribution of this paper. When the first versions of the VESC firmware with a built-in Lisp runtime system were released, there was some grumbling here and there about Lisp and its syntax, as expected. To our delight, most seemed very excited about a scripting language in the firmware. To the engineer and builder users, the choice of syntax seemed to matter less than the potential it opened up for them.
Read the full abstract

Machine Learning

Interpretable Multi-Hypersphere Deep Anomaly Detection for Open-set Supervised Anomaly Detection

Zhiji Yang, Fangyong Wang, Yue Li et al. (+2 more)

Multi-class open-set anomaly detection requires a model to characterize the normal acceptance domain formed by multiple heterogeneous subdistributions using only class-labeled samples from known normal classes, and to identify previously unseen anomalies at test time. Existing single-hypersphere methods cannot explicitly represent class-specific locations and acceptance ranges, while current multi-hypersphere or multi-class approaches do not fully integrate inter-class boundary constraints, learnable acceptance ranges, and interpretable decisions. To address these limitations, we propose Interpretable Multi-Hypersphere Deep Anomaly Detection (IMHD-AD). IMHD-AD constructs an independent hypersphere for each known normal class in a shared feature space. With target-inside and non-target-outside constraints, IMHD-AD embeds the class-specific hypersphere centers and radii directly into the final network layer and jointly optimizes them with the shared representation. The minimum signed boundary score across hyperspheres simultaneously determines open-set acceptance or rejection and provides a faithful geometric explanation of each decision. On MNIST, Fashion-MNIST, and CIFAR-10, IMHD-AD achieves the highest AUC in 28 of 30 open-set comparisons. A two-dimensional synthetic study further shows that model architecture must balance the compactness of known normal classes against the separability of unknown anomalies.
Read the full abstract

All authors Zhiji Yang, Fangyong Wang, Yue Li, Xianli Pan, Jianhua Zhao

Computer Vision

Compressing 3D Gaussian Splatting via Cross-Representation Priors

Yezheng Zhang, Huanxiong Liang, Chuqin Zhou et al. (+2 more)

3D Gaussian Splatting (3DGS) enables high-quality novel view synthesis but incurs high storage and transmission costs due to dense Gaussian primitives. Recent anchor-based compression reduces per-primitive redundancy, yet redundancy across anchors remains largely unexploited. We propose CRP-GS (Cross-Representation Priors for Gaussian Splatting), a rate-distortion optimized compression framework that leverages cross-representation priors to improve anchor-level entropy modeling. First, a Correspondence-Oriented Hierarchical Structure (COHS) organizes anchors by feature correspondence rather than spatial proximity, constructing root-leaf dependencies so that selected anchors can act as informative priors to conditionally encode others, yielding more accurate likelihood prediction and lower conditional entropy. Second, Shared Feature Aggregation (SFA) extracts globally shared features from a contextual hash grid and injects them into anchor representations, factoring out scene-consistent low-frequency information that would otherwise be redundantly embedded in individual anchors. Both modules are trained under a unified rate-distortion objective to balance bitrate reduction and rendering fidelity. Experiments across multiple benchmarks show that CRP-GS achieves a favorable overall rate-distortion trade-off, yielding around 30% average bitrate reduction compared to anchor-based baselines while maintaining comparable rendering quality.
Read the full abstract

All authors Yezheng Zhang, Huanxiong Liang, Chuqin Zhou, Guo Lu, Wenjun Zhang

Computer Vision

M3GA-Wild: A Large-Scale Dataset and Benchmark for Multi-Modal Multi-session Ground-to-Aerial Place Recognition in Forests

Ethan Griffiths, Maryam Haghighat, Simon Denman et al. (+2 more)

We present M3GA-Wild, the first benchmark for multi-modal, multi-session ground-to-aerial place recognition in forests. M3GA-Wild unifies and extends existing forest localisation datasets, providing a holistic benchmark with synchronised RGB imagery and LiDAR from ground traversals spanning 36 km, aligned high-resolution aerial imagery and multi-altitude LiDAR covering 370 hectares, and accurate geo-referenced 6-DoF poses for precise evaluation. M3GA-Wild captures diverse forest scenes with varying viewpoints, occlusion, and environmental conditions, enabling systematic evaluation of visual, LiDAR, cross-modal, and multi-modal methods. Baseline experiments show that LiDAR-based approaches significantly outperform vision-only methods under severe viewpoint differences, while current multi-modal fusion strategies yield limited gains due to poor cross-modal alignment. By pairing aerial RGB imagery with geo-referenced aerial LiDAR, M3GA-Wild also enables evaluation of foundation models for monocular depth estimation as a cheap source of 3D geometry from forest imagery, with initial experiments revealing shortfalls of current methods. These results highlight key challenges in cross-platform localisation, including modality misalignment and severe domain gaps. M3GA-Wild establishes a new benchmark to support research in robust multi-modal localisation and long-term autonomy in unstructured natural environments. The dataset and code will be available upon acceptance.
Read the full abstract

All authors Ethan Griffiths, Maryam Haghighat, Simon Denman, Clinton Fookes, Milad Ramezani

Numerical Analysis

NeSTR: Neural S-Transform Reconstruction for Gaussian SPDEs

Nacira Agram, Fred Espen Benth and Jan Rems

We introduce Neural $S$-Transform Reconstruction (NeSTR) for Gaussian stochastic partial differential equations (SPDEs). The learned object is not the random solution itself, but its $S$-Transform on a finite dimensional subspace of the white noise test space. For Wick type SPDEs, NeSTR yields a deterministic parametric PDE for the restricted $S$-Transform, and the stochastic solution is recovered from the Taylor coefficients of this transform at the origin. Thus the neural approximation is tied directly to the Wiener chaos expansion of the solution. The main novelty of NeSTR is an inverse $S$-Transform reconstruction principle combined with finite mode learning: we prove consistency of the recovered chaos coefficients, separate noise mode, chaos truncation, deterministic solver, and neural approximation errors, and give a deterministic learning formulation for the restricted transform. Numerical tests on an additive stochastic heat equation and a multiplicative Wick heat equation show that polynomial spectral representations in the transform variables provide stable access to higher order chaos coefficients. The resulting reconstruction separates the equation dependent deterministic coefficient fields from a reusable Gaussian–Hermite basis. Consequently, NeSTR learns an analytic generating function whose derivatives recover the Wiener chaos coefficients, rather than learning stochastic trajectories directly.
Read the full abstract

Systems and Control

Analysis and Design of Desaturation Short-Circuit Protection for SiC MOSFET Under High Switching Voltage Transients

Ravi Teja Pogulaguntla and Arun Karuppaswamy B

Short-circuit protection is vital for ensuring the safe operation of high-power-density power converters using Silicon Carbide (SiC) MOSFETs. However, adapting conventional desaturation (DESAT) protection, originally designed for Silicon (Si) IGBTs, is challenging due to the significantly shorter shortcircuit withstand time of SiC devices, typically less than 3{\mu}s. Nevertheless, DESAT protection can be adapted for SiC MOSFETs through certain circuit modifications. This article provides a comprehensive analysis and design guidelines for the three architectural variants of DESAT protection circuits, specifically for adjusting the short-circuit detection time suitable for SiC MOSFETs. Further, this article establishes a quantitative framework to evaluate the noise induced into the DESAT protection circuits during turn-off dv/dt switching transients. From this analysis, a design methodology and safe operating boundary are derived for each DESAT configuration, guiding the systematic selection of component values to balance dv/dt noise margin and short-circuit detection speed for SiC MOSFETs. Experimental results are presented to validate the analysis.
Read the full abstract

Human-Computer Interaction

Adaptive Scaffolding Needs Contingency: An AI Tutor That Escalates and Fades on What the Learner Does

Xinmeng Hou, Yuxuan Weng, Chin Hsien Yeh et al. (+5 more)

Coding assistants raise task performance, but learners plan and monitor less. Giving less away, the usual fix, conflates two things: how much work a system carries (cognitive load) and what the learner must decide before help arrives (metacognitive demand). Our principle, preserved metacognitive demand, holds the second constant and lets the first vary. CoMeT implements it: support rises when a learner fails at a decision point and fades on take-up. Within subjects, 131 adult learners used CoMeT, an unrestricted assistant and a question-only tutor on three Python tasks. CoMeT matched the question-only tutor's demand, delivered artifacts twice as often as the assistant, and frustrated learners less than the question-only tutor, with delegation and load unchanged. Learners often did not answer. Fading held when their turn addressed the decision under support, and CoMeT surrendered the full answer in one session in sixteen, against one in six for the question-only tutor.
Read the full abstract

All authors Xinmeng Hou, Yuxuan Weng, Chin Hsien Yeh, Ding Lin Lee, Lishan Zheng, Fang Li, Wuqi Wang, Yang Liu

Distributed Computing

Silent Failures Beyond the 32-Bit Index Range: A Differential Characterization of Large-Tensor Matrix Multiplication in PyTorch's MPS Backend

Junichiro Niimi

Apple Silicon machines with large unified memory make it possible to hold large tensors on a desktop GPU. However, we found that PyTorch's Metal Performance Shaders (MPS) backend silently returns wrong results for batched matrix multiplication with more than $2^{32}$ elements. torch bmm, including its wrappers matmul and eager attention, returns relative errors above 1 without an exception or a warning in every PyTorch release tested (2.4.1 to 2.14.0). We sweep bmm over dtypes, memory layouts, shapes and batch sizes around $2^{31}$ and $2^{32}$ elements, and judge every result against a float64 computation on the CPU. Three rules account for every outcome on 2.14.0. When the output exceeds $2^{32}$ elements and an operand is a transposed view, the entire output is wrong and equals a computation that ignores that operand's strides. Otherwise, a view with at least $2^{31}$ elements raises an exception, and a contiguous input above $2^{32}$ elements makes exactly the batches beyond that point wrong, equal to a computation whose index wraps at $2^{32}$. A slightly larger problem can thus turn an explicit error into a silent failure. The rules extend to the backward pass, where a correct forward pass can return silently wrong gradients. A second machine with another chip, under two macOS versions, reproduces all 6156 results, including the wrong values, and the same sweeps on an NVIDIA A100 are correct in all 2530 runs. In a public sentiment classifier, one oversized batch corrupts a third of the outputs, which collapse onto one class. All findings come from observable behavior, without access to the backend's closed-source kernels; we release the harness, raw results and a guard that stops any MPS operation touching $2^{32}$ or more elements at jniimi/mps-silent-failures (https://github.com/jniimi/mps-silent-failures).
Read the full abstract

Machine Learning

On attention heads and bilinear forms

Andrew O'Desky

We study the symmetric and antisymmetric parts of bilinear forms in the attention heads of trained large language models. We introduce an orthogonally invariant profile map from real bilinear forms to a three-dimensional simplex and observe that profiles of trained bilinear forms accumulate near profiles of rank-one bilinear forms. We prove that the symmetric part of a bilinear form in an attention head is the sum of a hyperbolic form and a zero form for a Zariski-dense subset of query-key matrices.
Read the full abstract

Computation and Language

Rethinking Pivot Programming Languages in Code Language Models

Andor Diera, Lukas Galke Poech and Matthias Tichy

Multilingual code language models transfer skills across programming languages (PLs), but whether any PL occupies a privileged pivot position remains contested: geometric analyses point to C-family languages and Go, while behavioral evidence highlights Python. We revisit this question under controls for representational anisotropy and length variation across PLs, two confounds that compromise prior cosine-based analyses. Across three code models on multilingual competitive-programming data, we study three views of cross-PL organization: pairwise PL geometry, PL-English alignment, and pivoted retrieval through candidate PL representation spaces. The results are relation-dependent. Code-code geometry reveals structured language regions but no universal center; code-English alignment favors high-level scripting languages; and pivoted retrieval favors different intermediate spaces for code-to-code and English-to-code transfer. These findings suggest that Python's special role is better understood as English-facing affinity than as universal geometric centrality.
Read the full abstract

Artificial Intelligence

OptiSkill: A Hierarchical and Evolving SkillBank for LLM-Based Optimization Modeling

Ruiqing Zhao, Rui Liu, Yuan Zuo et al. (+3 more)

Automated operations research (OR) modeling requires LLMs to translate natural-language decision problems into correct mathematical programs. Existing methods can improve individual formulations, but they often solve problems in isolation, retaining little reusable experience and repeating similar formulation errors. Prior memory-based approaches store examples, thoughts, or insights as references, while OR modeling requires reusable formulation skills that transfer across problem narratives and guide concrete modeling decisions. We propose OptiSkill, a skill-augmented framework that builds a hierarchical and evolving SkillBank for LLM-based OR modeling. SkillBank stores solver-verified experience as reusable skills, with Global Strategies for problem-level formulation skeletons and Step Experiences for local error-prevention rules. It is further refined through stable batch-level test-time evolution, where candidate skills are incorporated only after validation. Experiments on eight OR modeling benchmarks show that OptiSkill improves formulation accuracy across LLM backbones, outperforms strong agentic baselines, and gains further by expanding SkillBank coverage and reliability. Code and data are available at https://github.com/rachhhhing/OptiSkill
Read the full abstract

All authors Ruiqing Zhao, Rui Liu, Yuan Zuo, Huarong Zhang, Xiao Han, Junjie Wu

Systems and Control

On a Closed-Loop Controller for the Coevolutionary Model of Actions and Opinions via Broadcasting Information

Roberta Raineri, Mengbin Ye and Lorenzo Zino

We deal with controlling a complex social network in which agents have actions and opinions that coevolve, mutually influencing one another. We consider an input consisting in broadcasting information to a target set of agents with the objective of steering the population, initially at a consensus, to a different consensus. For a constant input, we derive a monotone convergence result, building on which we design an algorithm that determines whether a target set is sufficient to achieve the objective and an effective heuristic to optimize the target set. Then, we introduce a feedback control law that, using information on the state of the system, dynamically revises the target set, reducing the effort needed to achieve the objective while guaranteeing convergence to the desired consensus state.
Read the full abstract

Machine Learning

LPINNs: First-Layer Gated Localization for Physics-Informed Neural Networks

Lakshay Chawla and Hardik Jain

Physics-informed neural networks (PINNs) use one shared representation over the computational domain, which can become difficult to optimize on long domains and for high-order operators. We study a minimal alternative: multiply the first hidden activation of an otherwise unchanged dense PINN by input-dependent localization functions, giving first-layer units receptive fields without partitioning the domain or adding interface losses. We screen 13 families of localization functions, in up to three parameterizations each, on a nonlinear harmonic oscillator (HO), a heat equation on a long spatial interval, and a manufactured four-dimensional (4D) fourth-order problem, with ten paired seeds throughout. Three configurations give large reductions in solution error at matched budgets: (i) Fixed Gaussian localization functions on the $2\pi$ HO domain cut mean solution RMSE from $4.8369\times10^{-1}$ to $8.83\times10^{-3}$ at 3k epochs. (ii) The inverse-quadratic family with learnable centers and widths cuts it from $2.896\times10^{-1}$ to $3.06\times10^{-2}$ on the $8\pi$ heat domain at 10k epochs. (iii) Fixed bump localization functions cut it from $1.75947\times10^{1}$ to $2.260\times10^{-1}$ on the $4\pi$ 4D domain at 10k epochs. Every paired seed improves in these three comparisons. The screen also shows that the mechanism is not a free win: on HO only 2 of 13 families beat the baseline, and 10 of the remaining 11 are 9 to 23 times worse; on 4D four families are non-finite and five are more than three orders of magnitude worse than the baseline. The inverse-quadratic family is the only one that beats the baseline on all three equations. Overall, these results show that first-layer localization can provide measurable improvements to baseline PINNs on long-domain and high-order problems.
Read the full abstract

Robotics

Connectivity-Aware Exploration of Robotic Grasp Spaces

Maksim A Kazanskii

Robotic grasping is typically formulated as the problem of identifying successful actions from a space of candidate grasp poses. However, the organization of successful actions within this space has received less attention. We study the multiscale structure of viable robotic grasps in $SE(3)$ and investigate whether this structure can be exploited for more efficient exploration. Using a large-scale grasp dataset, we show that successful grasp sets exhibit heterogeneous and reproducible connectivity structure across objects. We then introduce a connectivity-aware sampling strategy that incrementally explores the currently observed grasp space by prioritizing potential bridges between components, structural frontiers, boundary extensions, and geometric novelty. In controlled reconstruction experiments, the method recovers the connectivity structure of successful grasp sets substantially more efficiently than random sampling and farthest-point sampling. We further evaluate whether connectivity acquired under hidden grasp viability can improve subsequent grasp discovery, and whether structural experience from previously explored objects can be retrieved and transferred to unseen objects. These results suggest that the spatial organization of viable actions provides information relevant to grasp-space exploration beyond the viability of individual candidate actions. More broadly, they motivate structure-aware exploration as a means of exploiting the geometry of viable action spaces in robotic manipulation.
Read the full abstract

Cryptography and Security

Dual-Locking Learned AI Models: A PIN-Based Sparse QIM Watermarking and Adaptive Index Permutation Approach

Iva Vasic, Jesús Muñoz-Cádiz and Bata Vasic

We present a dual-locking method for securing trained neural networks that combines key-driven index permutation with PIN-based watermarking based on Sparse Quantization Index Modulation (QIM). Cryptographic randomness is introduced by independently applying a uniform random permutation to each row of adaptively selected index vectors. A robust blind binary watermark is then embedded into the bias coefficients by modulating their quantized values, binding the network to a user-defined Personal Identification Number (PIN). Without the correct key, the network retains its architecture but becomes functionally impaired due to disrupted internal representations. Inverse permutation fully restores the original model accuracy, while the embedded watermark remains imperceptible and enables blind verification of key association and model authorship. To improve both locking effectiveness and recoverability, an adaptive key selection strategy redistributes high-magnitude weights to low-sensitivity positions and vice versa, increasing degradation in the locked state while preserving full recovery. Experiments on MNIST, CIFAR-10/100, and ImageNet-1K using fully connected networks, ResNet CNNs, and transformer architectures show that locking reduces accuracy below 10%, and even below 0.5% for CNNs, while the correct key fully restores performance. The watermark introduces no measurable accuracy degradation and reliably authenticates ownership. Analysis of embedding distributions across CNNs and transformers further indicates potential diagnostic value for identifying undertrained or suboptimally designed models. The proposed approach therefore provides simultaneous model protection, recovery, and ownership verification.
Read the full abstract

Distributed Computing

DeepSeek Elastic Compute (DSec): A Sandbox Infrastructure for Effective Agentic Training at Scale

Jialiang Huang, Hongxuan Tang, Jingchang Chen et al. (+128 more)

Large-scale agentic training and evaluation with large language models (LLMs) rely on isolated, stateful execution environments in which models inspect repositories, invoke tools, execute commands, and interact with task-specific services. These workloads create sandboxes in large bursts, span heterogeneous functionality and isolation requirements, retain state across long interactions, and draw from large image corpora with limited reuse. Supporting them therefore requires an elastic execution platform rather than a single sandbox runtime. This report presents DeepSeek Elastic Compute (DSec), a production sandbox platform that exposes FnCall, container, microVM, and full-VM sandbox backends through a unified SDK. DSec coordinates placement and lifecycle management across the cluster, composes environments from independently versioned layers, combines memory sharing, reclamation, and CPU scheduling for high-density execution, and loads image data on demand from Fire-Flyer File System (3FS), a cluster-wide distributed filesystem. DSec is co-designed with the reinforcement learning (RL) framework, decouples stateful rollout execution from preemptible GPU training, coordinates sandbox lifecycle with training to preserve rollout state while reclaiming idle resources, and mitigates agent misbehavior such as reward hacking. A single production-scale unit of DSec spans around 160 nodes, serving about 3 million sandboxes per day; in production, it supports over 380,000 concurrent sandboxes and sustains over 5,000 sandbox creations per second. Our evaluation and deployment experience show that these mechanisms reduce environment setup and image-distribution overhead, improve memory efficiency, and preserve latency-sensitive performance under high-density overcommit.
Read the full abstract

All authors Jialiang Huang, Hongxuan Tang, Jingchang Chen, Yuxuan Liu, Yixiao Chen, Yuan Cheng, Yi Tao, Jingli Zhou, Yupeng Chen, Haoyu Chen, Jiarui Wang, Shengkai Lin, Chuqi Zhang, Bryan Lee Teng, Lian Guo, Zhe Fu, Wenjun Gao, Yisong Wang, Liang Zhao, Zehao Wang, Ziwei Xie, Yongqiang Guo, Peixin Cong, Ziyi Gao, Shuiping Yu, Hanwei Xu, Zuofan Wu, Zhizhou Ren, Yuyang Zhou, Bowei Zhang, Zhihuan Huang, Qihao Zhu, Lei Wang, Tianle Lin, Han Yu, Jiewen Hu, Dejian Yang, Shuo Yang, Shanghao Lu, Shaoyuan Chen, Junjie Qiu, Zhangli Sha, Yinmin Zhong, Yongtong Wu, Shiyu Wang, Wei Liu, Bingzheng Xu, Longhao Chen, Qiushi Du, Yuzhen Huang, Shirong Ma, Yaohui Wang, Mingshu Chen, Tongrui Xiong, Y.C. Yan, Haowen Luo, Haofen Liang, Xiaokang Zhang, Weihao Zeng, Runxin Xu, Peiyi Wang, Jinhua Zhu, Ruoyu Zhang, Wenkai Yang, Qi Tang, Jiping Yu, Tian Ye, Ruizhe Pan, Honghui Ding, Xiaodong Liu, Lingxiao Luo, Zhihong Shao, Yuhan Wu, Jibai Lu, Wen Liu, Haoling Zhang, Jingcheng Hu, Yaoyang Ye, Chaofan Lin, Zhaochen Zhang, Jianan Tong, Hengxu Wu, Zhihao Li, Yicheng Wang, Luyao Wang, Yuzhuo Bai, Lingyue Fu, Ruifan Xu, Y.Z. Wang, Zonglin Li, Mingqi Wei, Haiyang Shen, Chengyuan Zhang, Chao Jin, Zili Zhang, R.H. Yang, Xinbo Xu, Jian Zhou, Ruidong Zhu, Yuzhe Guo, Zelun Pan, Shaoheng Nie, Erhang Li, Shuhan Lin, Zheng Liu, Anshuo Chen, Zilong Lyu, Sinuo Cao, Rui Yu, Chuhao Wang, Junyi Guo, Junxiao Song, Kaifeng Chen, Menghao Ye, Junxian Li, Di Wu, Haiyang Ma, Yilun Wang, Haoran Yang, Yizai Cai, Shichun Liu, Yiping Wang, Junbo Sun, Shicheng Xu, Xiao Bi, Ying He, Yichao Zhang, Mingxing Zhang, Liyue Zhang, Panpan Huang, Wenfeng Liang

Machine Learning

Beyond Similarity: Coverage-Aware Prompt Selection for Time Series Forecasting with LLMs

Daeun Ji, Minkyoung Kim, Dongkuk Kim et al. (+3 more)

Similarity-based retrieval is the dominant rule for conditioning large language models (LLMs) in in-context learning, retrieval-augmented generation, and prompt-based time series forecasting. The rule concentrates on near-duplicate candidates, an issue that has motivated diversity-aware retrieval but remains unexamined in other retrieval-conditioned pipelines. We study this issue using prompt-based time series forecasting as a test bed, where a learned prompt pool is retrieved by similarity. Dominant methods in this setting retrieve top-K entries by cosine similarity without redundancy control, producing a bias toward dominant temporal patterns while overlooking rare but informative events. We propose CASP-LLM, a coverage-aware semantic prompting framework that addresses this prompt selection bias by combining usage-tracking and saturating-gate techniques into a coverage regularizer that adds no learnable parameters. On six long-term benchmarks and the M4 short-term benchmark, CASP-LLM matches or improves on similarity-based LLM forecasters on most dataset-horizon settings, with the exceptions of Electricity, M4-Monthly, and the few-shot long-horizon setting. A controlled study locates the failure mode at the cross-batch usage level rather than per-retrieval redundancy: within-retrieval diversification such as MMR does not help, whereas regularizing anchor usage across training does.
Read the full abstract

All authors Daeun Ji, Minkyoung Kim, Dongkuk Kim, Yohan Lee, Beomsoo Kim, Beakcheol Jang

Robotics

A Horizon-slicing Approach to Minimum Obstacle Displacement Planning for Robot Navigation

Antony Thomas, Giulio Ferro, Fulvio Mastrogiovanni et al. (+2 more)

In this paper, we investigate the Minimum Obstacle Displacement Planning problem from a robot motion planning perspective. The problem involves determining a feasible path to a goal location by displacing movable obstacles when no collision-free path initially exists. We show that this problem is computationally challenging and, in particular, NP-hard when obstacles are modeled as polygons in the plane. Besides an exact formulation of the minimum obstacle displacement problem generalizing other problems in the literature, and the associated optimal solution, this paper proposes an approximate solution that is less intensive from a computational standpoint, and differs from the optimal solution by a fraction of the optimal cost, being able to trade-off between path length and amount of obstacle displacements.
Read the full abstract

All authors Antony Thomas, Giulio Ferro, Fulvio Mastrogiovanni, Michela Robba, Marco Baglietto