Artificial Intelligence
Papers
Fresh from arXiv, typeset properly. Highlighted words trade on the market.
All of arXiv
Saturday 29 August
Predicting future organ dysfunction in Intensive Care Unit (ICU) patients is critical for early clinical intervention, yet existing machine learning approaches have largely treated the Sequential Organ Failure Assessment (SOFA) score as an input to binary mortality prediction rather than as a continuous clinical outcome in its own right. We investigate the extent to which a Temporal Convolutional Net work (TCN) can predict next-day SOFA scores from multivariate ICU time-series data extracted from MIMIC-IV, characterise the relative contribution of each organ system to total SOFA variance and deterioration, and identify distinct trajectory patterns across ICU stays. A residual TCN trained on three-day sliding windows achieved a five-fold cross-validation R2 of 0.740 +- 0.013 and MAE of 1.431 +- 0.022, outperforming a naive persistence baseline on RMSE and R2. SHAP interpretability analysis revealed that the model functions primarily as a severity-anchoring mechanism rather than a true sequence model, with predictions dominated almost entirely by the most recent observation day. Cardiovascular dysfunction emerged as the strongest discriminator of both cross-sectional severity and acute deterioration, and unsupervised trajectory clustering identified two clinically meaningful phenotypes, an improving group (58.9%) and a persistently severe group (41.1%), differentiated by cardiovascular, hepatic, coagulation, and renal involvement. We conclude that TCNs can extract meaningful predictive signal from ICU physiological data, but that short input windows and complete-case selection bias currently limit their clinical utility, motivating future work on longer input horizons, alternative missing-data strategies, and external validation.
Optimization and Control
Preconditioned Hyperbolic Smoothing Modified L-BFGS Algorithms for Finite Minimax Problems
This paper proposes a preconditioned ordinary hyperbolic smoothing modified L-BFGS framework for finite minimax problems. To mitigate the deterioration of Euclidean conditioning as the smoothing parameter decreases, a problem-derived hyperbolic majorization preconditioner is incorporated into the initial L-BFGS metric; it combines component-gradient Lipschitz curvature with the curvature induced by the hyperbolic smoothing Jacobian, is uniformly positive definite, and provably majorizes the Hessian of the smoothed objective. To exploit strict negative secant curvature, three correction strategies are introduced—direct sign reflection, a Euclidean nearest-point correction, and a $B_k^{-1}$-metric nearest-point correction—for which explicit formulas are derived and Clarke-stationary accumulation points are obtained for the idealized continuation. For a fixed smoothing parameter, Q-linear convergence of the HMLBFGS objective values is obtained under a Polyak–Lojasiewicz condition, local strong convexity further yields R-linear convergence of the iterates, and the corresponding full-memory modified BFGS methods attain local Q-superlinear convergence under the usual unit-step assumptions. Numerical experiments demonstrate the effectiveness of the proposed preconditioned methods and their numerical advantages over the selected comparison methods.
Read the full abstract
Optimization and Control
Regularized Coupling Maps on Matrix Polytopes: From Assignment to Graph Matching and Optimal Transport
Three classical problems-linear assignment, graph matching, and optimal transport-share a common structure: they seek an optimal matrix-valued coupling on a matrix polytope. We study regularized coupling maps that encompass two families of regularization: entropy/KL regularization, which exponentiates the input score matrix and projects it onto the relevant matrix polytope via the KL-projection; and the Frobenius/Euclidean regularization, which performs the Euclidean projection of the scaled input matrix onto the same polytope. We trace how these two geometries yield complementary algorithmic tools for graph matching (CSGO, ASM, FRAM) and optimal transport (IP-EOT, PSN). The regularized coupling map serves as a direction generator in graph matching and as a solution representation in optimal transport, with the regularizer determining the geometry and the outer task determining the precision standard.
Read the full abstract
Differential Geometry
On Hypersurfaces of Space Forms That Are Gradient Yamabe Solitons
This paper studies nontrivial Yamabe gradient solitons occurring as complete immersed hypersurfaces with constant scalar curvature in space forms $\left(\overline{M}^n(C), \overline{g}\right)$. For solitons with non-vanishing gradients, we establish a structural characterization of the soliton by analyzing a regular level set $\Sigma$ of the soliton function. In particular, imposing a suitable condition on the traceless part of the second fundamental form, $\Phi^{\Sigma}_{\overline{M}}$ of $\Sigma$ as a submanifold of the space form, we prove that the soliton either decomposes as a Riemannian product of $\mathbb{R}$ and a totally umbilical submanifold of the space form, or satisfies a sharp lower bound on $\sup\limits_\Sigma\left|\Phi^{\Sigma}_{\overline{M}}\right|$. Furthermore, we show that the lower bound is attained if, and only if, the soliton is isometric to a Riemannian product of $\mathbb{R}$ and a parallel submanifold of the space form.
Read the full abstract
Systems and Control
A Catalog of Probability Generating Functionals for Multitarget Tracking and Data Assignment Problems
This paper studies the class of Bayesian tracking filters for which the probability generating functional of the joint target-measurement process can be derived from the statistical assumptions that define the problem. The class includes filters for labeled and unlabeled targets, as well as hybrid filters in which both labeled and unlabeled targets are present. New results include the probability generating functional for interval filtering of target trajectories for both labeled an unlabeled targets, and a novel method for computing the probability generating function of measurement-to-target assignment probabilities. Probability generating functionals give exact expressions for calculating the importance weights in particle filter implementations. Low computational complexity approximations to the particle weights are derived from the probability generating functionals via the saddle point method.
Read the full abstract
Machine Learning
When Do Larger Batches Help Scale LLM Reinforcement Learning?
Larger batches reduce the variance of stochastic gradients per update and are therefore often expected to accelerate training. Yet whether this statistical benefit translates into lower wall-clock time-to-target remains unclear, because each update consumes more samples and may take longer to execute. We study this tradeoff in reinforcement learning for large language models. We separate its algorithmic and systems effects by comparing learning and execution along their natural axes. At the algorithmic level, we compare configurations at equal cumulative sample counts while retuning batch-dependent hyperparameters. Over a bounded range of batch sizes, this procedure yields an approximately batch-size-invariant family whose members follow similar sample-indexed learning trajectories. At the systems level, we exploit the computational asymmetry between rollout generation and training: autoregressive generation is often memory-bandwidth-bound at low concurrency, whereas training work scales approximately with the number of processed tokens. Combining these two views yields a direct decision rule: a larger-batch configuration reduces time-to-target only when its throughput gain exceeds its samples-to-target penalty. Experiments with GRPO and PPO support both sides of this decomposition. At the algorithmic level, square-root learning-rate scaling with Adam produces approximately batch-size-invariant learning curves over a bounded range of batch sizes. At the systems level, larger batches improve generation throughput by up to 2.29x on fixed hardware. In GRPO, combining higher throughput with learning-rate retuning reduces time-to-target by up to 29%, whereas increasing the batch without retuning is slower despite its higher throughput.
Read the full abstract
General Relativity and Quantum Cosmology
Automated identification and subtraction of gravitational-wave glitches using boundary refinement
Transient noise artifacts, or glitches, in gravitational wave strain data elevate the false alarm rate of astrophysical searches and degrade parameter estimation when overlapping a signal. No method previously identified a glitch's time boundary: existing detection and classification tools flag and label glitches without resolving their extent, forcing subtraction to run over padded windows that cost time and erase signal beyond the glitch itself. We present three boundary identification methods, AMPS (Amplitude-based Multi-glitch Pulse Segmenter), FLARE (Fitness-based Localization And Refinement Extractor), and CRISP (connected-region identification via spectrogram power), paired with three subtraction techniques, adaptive spline fitting, wavelet shrinkage, and their combination, across five glitches from the GravitySpy database spanning Advanced LIGO's first three observing runs. AMPS sets a boundary from a robust amplitude threshold, FLARE from the best fitness of a segmented spline fit, and CRISP from spectrogram power. Injecting a chirp signal on four broadband glitches, the combined technique recovers 95 to 97 percent of the injected signal-to-noise ratio, against roughly 64 percent for wavelet shrinkage alone. CRISP gives the most uniform boundary width across glitches; AMPS and CRISP both identify a boundary in a fraction of a second, two to three orders of magnitude faster than FLARE. For the glitch overlapping GW170817, a low-frequency residual persists under the default boundary, a deliberate tradeoff against removing signal power. Scattered light glitches remain untested and are the primary direction for future work.
Read the full abstract
Functional Analysis
Numerical range and invariant subspaces of weighted composition operators on the Hardy space of Dirichlet series
In this paper, we study various results on numerical range of weighted composition operators on Hardy space of Dirichlet series. Certain conditions are provided under which the numerical range contains zero and some circular or elliptic disks. With the help of invariant subspaces of the operator, we also determine the numerical range of reductive weighted composition operators on this space. Additionally, we discuss some results on the Davis-Wielandt shell of composition operators with some natural questions that arise from our findings.
Read the full abstract
General Relativity and Quantum Cosmology
Optimizing Superconducting Microwave Cavities for Gravitational Wave Sensing
Superconducting microwave cavities loaded with radio frequency fields are a powerful tool to search for weak forces or electromagnetic perturbations due to new physics. One source of such signals can be high-frequency gravitational waves emitted from cosmological or unknown astrophysical events. However, large sensitivity improvements are still necessary to reach the parameter space motivated by theoretical models. In this work, we present a formalism to guide the design of such detectors across a broad range of frequencies and signal forms. By incorporating the interdependencies of all relevant parameters and the back-action of the electromagnetic fields on the cavity structure, we describe how figures of merit can be derived for a broad class of relevant experimental setups. Using a two-dimensional model, we demonstrate this formalism and present examples for optimized cavity geometries. We find that the choice of geometry alone can increase the signal-to-noise power ratio by an order of magnitude compared to an existing prototype of the same size. We also show that different physics goals lead to different optimal cavities, thus proving the need to consider such figures of merit at an early stage when designing a new detector.
Read the full abstract
Human-Computer Interaction
Measuring the "Interaction Gap" in Drama Therapy with AI
Generative AI is increasingly being introduced into expressive arts therapy, where it is often credited with offering a non-judgmental environment that supports psychological safety. Existing HCI work has largely positioned AI as a co-creative material or as a bridge/mediator into human-led care. This paper explores a different position. When a patient performs the same drama therapy task with an AI partner and with a human partner, the resulting self-presentations tend to differ in patterned ways. We propose treating this difference, the Interaction Gap, as a diagnostic lens within drama therapy. Rather than asking which context elicits a truer self, the lens reads the difference between the two performances as information about the social pressures shaping self-expression in each context. We sketch a starting point for task design and measurement signals grounded in drama therapy's existing use of role and aesthetic distance, and raise provocations for workshop discussion: the observer effect and privacy paradox that measurement introduces, and the question of whose lens the gap is.
Read the full abstract
Artificial Intelligence
Accelerating Unified Multimodal Models with Core-Expansion Routing and Unified Computation Scheduling
Unified multimodal models jointly support understanding and generation, but incur substantial redundant computation across tokens, layers, and generation timesteps. Through token-importance probing, we identify an asymmetric core-expansion structure: understanding exhibits a stable importance component, while generation largely shares this component but requires progress-dependent corrections. We therefore propose CE-Router, which uses a task-shared core scorer and progress-conditioned generation expansions, optimized through generation decomposition and cross-task core alignment. At inference, CE-Router compacts token computation and supplies a learned routing signal to Unified Computation Scheduling, which coordinates layer skipping, FFN pruning, diffusion-head cache reuse, and denoising-step early exit. Experiments on two representative UMM architectures demonstrate consistent quality–efficiency improvements across both tasks, retaining 98.03% of dense understanding performance with a 1.93$\times$ end-to-end inference speedup.
Read the full abstract
Information Retrieval
Database-Augmented RAG for Automated Repair of REST API Misuses
Many Internet of Things (IoT) services provide Representational State Transfer (REST) APIs, which require client developers to implement applications that conform to the corresponding API specifications. When client programs contain API misuse, developers debug them based on error responses. However, such responses are often insufficient for identifying the root cause, requiring developers to repeatedly communicate with the server. Retrieval-Augmented Generation (RAG) is a promising approach for providing large language models (LLMs) with external knowledge. However, in automated repair of REST API misuses, it remains unclear how specifications should be stored in a RAG database. This study evaluates how different configurations for organizing API specifications affect RAG-based repair of REST API misuse. We constructed 11 RAG configurations with different database structures and compared their repair rates with a baseline method. For evaluation, we used REST API misuse cases collected from real-world repositories. The results show that, in the studied datasets, the baseline method achieved a repair rate of 54.3%, whereas a RAG-based method using four databases achieved a maximum repair rate of 88.6%. These results indicate that organizing specifications according to version and content type can be an effective design choice for RAG-based REST API misuse repair.
Read the full abstract
Computer Vision
AOI-Net: Structural Face AOI-Guided Eye-Gaze Track Representation Learning for Autism Spectrum Disorder Detection
Eye-movement tracking has emerged as a promising non-invasive approach to Autism Spectrum Disorder (ASD) screening, with systematic differences in attentional allocation and revisit behaviors observed during socially interactive tasks. Existing computational methods typically characterize eye-movements using discrete gaze trajectories and fixation events, yielding representations dominated by short-range temporal dynamics and limiting models that primarily emphasize long-range dependencies. Meanwhile, gaze behavior is naturally organized across semantically meaningful Areas of Interest (AOIs), whose attention allocation and transitions provide important structural cues, yet their relationships are rarely modeled explicitly. To address these limitations, we propose a structural face AOI-guided Eye-Gaze Track Network (AOI-Net) that jointly models short-term temporal dynamics and AOI-level structural organization. A network gating mechanism adaptively integrates the complementary temporal and structural representations according to their contributions to gaze-behavior characterization. To mitigate the pronounced class imbalance commonly encountered between individuals with ASD and Typically Developing (TD) participants in clinical datasets, class-distribution-aware learning is further employed to facilitate discriminative embedding learning under skewed class distributions. Experiments on a unique and large-scale clinical eye-tracking database comprising eight stimulus subsets and more than 1,300 participants show that AOI-Net consistently outperforms state-of-the-art methods. The proposed framework also enables interpretable gaze-behavior modeling and provides a practical basis for scalable AI-driven ASD screening in real-world healthcare. The code is available at https://github.com/Zhanpei-ai/CIM-AOI-Net/tree/main/Code
Read the full abstract
Number Theory
Counterexamples to O'Neil's Period-Index Problem on Elliptic Curves
Let $E/K$ be an elliptic curve over a number field $K$, and let $[C]\in H^1(K,E(\overline{K}))$ be a homogeneous space under $E$. Suppose that $C$ has period $n$ and index $d$. O'Neil asked whether one can always choose a lift of $[C]$ to $H^1(K,E[n])$ whose period-index obstruction has order exactly $d/n$. We give a negative answer to this question by constructing an explicit family of examples.Taking $K=\mathbb{Q}(\zeta_8)$, we prove that there exist infinitely many pairwise non-isomorphic elliptic curves $E/K$ such that each $E$ admits infinitely many homogeneous spaces $[C]\in H^1(K,E(\overline{K}))$ with period $8$ and index $16$, but the period-index obstruction of every lift of $[C]$ to $H^1(K,E[8])$ has order exactly $8$.
Read the full abstract
Computer Vision
PERSIST: Persistent-State Discrimination for Shot Boundary Detection
Shot boundary detection (SBD) is widely treated as the localisation of local visual discontinuities, yet many false positives such as hand-held shake, illumination flicker, motion blur, occlusion, and damaged archival material produce equally sharp local change without introducing a new shot. We reformulate SBD as boundary semantic discrimination: a frame is favoured as a boundary only when its local change evidence is accompanied by a persistent update of the video's latent temporal state, rather than a transient excursion that returns to the surrounding trend. This persistence test is operationalised with a continuous latent state from a FiLM-conditioned sinusoidal representation network and a structured discriminator that combines three semantic cues, local change, transient impulse, and return-to-trend, into a single interpretable per-frame signal over a dual-rate temporal backbone. The resulting framework, PERSIST, turns every decision into an inspectable one: the persistence criterion is trained into the classifier, its per-frame effect stays readable from the gate triple, and its learned latent state is measurably boundary-discriminative. On a 2,727-video per-subtype diagnostic it removes 33-80% of flash, text-overlay, and archival false positives relative to an identically trained cue detector, and at matched true-transition recall it roughly halves TransNetV2's pseudo-event false positives on that diagnostic and cuts its false positives on ClipShots footage by about a quarter, while preserving recall. It does so while reaching parity with the strongest public detector across online, broadcast, short-form, and historical-archive transfer evaluations, under markedly stricter training: it learns from ClipShots real transitions only, whereas the anchor draws on additional corpora whose transitions are 85% synthetic. Code is available at https://github.com/linty5/PERSIST.
Read the full abstract
Artificial Intelligence
MMPCBench: Benchmarking Multimodal Large Language Models on Proactive Critique of Flawed Inputs
As Multimodal Large Language Models (MLLMs) evolve into sophisticated interactive assistants, their reliability depends not only on following instructions but also on validating them. We define Proactive Critique as the model's autonomous ability to identify, analyze and fix faulty user inputs without extra prompts. However, evaluations mainly test models under ideal circumstances or simple refusal behaviors, largely ignoring active error processing. To fill this gap, we propose MMPCBench, a comprehensive framework for evaluating MLLMs' proactive critique competence. It features a fine-grained taxonomy of 4 primary error types spanning 12 subcategories, ranging from cross-modal contradictions to missing visual premises. We adopt a hierarchical evaluation protocol to measure models' error detection, diagnosis and resolution performance, and apply alignment-aware metrics to assess the coherence between internal reasoning and final responses. Tests on 14 mainstream MLLMs show obvious weaknesses in proactive critique, especially in dealing with subtle visual anomalies. Notably, we identify a pervasive "consistency gap": reasoning models can often correctly identify and analyze errors during internal reasoning yet suppress these valid insights in final outputs to prioritize response compliance. The code and data is available at https://github.com/ALIENS32/MMPCBench.
Read the full abstract
Computer Vision
3D-MRL: Nested Multimodal 3D Representations via Matryoshka Representation Learning
Vision-Language Models align point clouds with image and text embeddings, enabling zero-shot recognition, retrieval, and open-vocabulary understanding of 3D shapes. Existing multimodal 3D pre-training methods produce fixed-dimensional embeddings, requiring separate models for different computational budgets. We propose 3D Matryoshka Representation Learning (3D-MRL), a multimodal 3D pre-training framework based on Matryoshka Representation Learning. 3D-MRL learns nested 3D representations by aligning point clouds with frozen CLIP image and text embeddings while applying contrastive supervision across multiple embedding dimensions. The Matryoshka objective is applied only to the 3D encoder, allowing a single model to produce representations at different dimensionalities without retraining. Experiments on the Objaverse-LVIS, ModelNet40, and ScanNet datasets show that 3D-MRL achieves competitive performance on zero-shot and few-shot 3D recognition tasks. In addition, the learned representations support retrieval across different embedding dimensions within a single model. On Objaverse-LVIS, 3D-MRL improves Top-1 accuracy from 46.8% to 50.9%. Retrieval experiments further show that different embedding dimensions yield varying levels of semantic and geometric specificity.
Read the full abstract
Computation and Language
Cloud and On-Premises Deployment of Uzbek Legal RAG via Targeted Retriever Fine-Tuning
Deploying large language models for legal question answering raises challenges that general-purpose leaderboards do not capture, particularly for low-resource languages and under hard operational constraints. We report on building and operating a retrieval-augmented (RAG) legal assistant for Uzbek that must run in two regimes: a managed cloud service that maximizes answer quality within a per-token cost ceiling, and an on-premises deployment for clients whose legal data may not leave their infrastructure, restricting us to open-weight models on limited local hardware under latency constraints. Because no evaluation existed for this setting, we build two domain benchmarks: a retrieval benchmark of 178 expert-annotated legal queries with gold provision spans, and an end-to-end benchmark of 504 expert-curated question–answer pairs scored by an LLM judge whose ratings we validate against human judgments and against an independent-family judge. Applying these benchmarks under each regime, we find the open-versus-proprietary gap is small and cheaply closed by fine-tuning. Therefore, we train UTE-1, which is a state-of-the-art text embedder among open models for Uzbek. We also demonstrate that closing the performance gap via fine-tuning is both impractical due to the intensive hardware demands of long-context legal Q&A and unnecessary, given that legal acts change frequently. We support this by reporting a negative result from a QLoRA experiment. We distill practical guidance for similar deployments, drawn from a system serving real users in production. We release our benchmarks, evaluation code and the fine-tuned embedder (UTE-1) \href{https://metric-ai-lab.github.io/Uzbek-Legal-RAG/}{at this https URL} to support future work on low-resource legal NLP.
Read the full abstract
Cryptography and Security
A Multi-Month Study of Git Commit Signing
Git commit signing, introduced in 2012, is one mechanism for establishing commit provenance in software supply chains, yet developer-controlled adoption remains rare and developers' experiences using it are understudied. To examine this experience, we conducted a three-month study with senior undergraduate and graduate computer science students (n = 22), whom we treat as proxies for junior developers. Participants configured commit signing independently, used it across four coursework projects, extended it to a second device, examined an external repository containing anomalous commits, and answered security-reasoning prompts. We found that while almost all participants successfully signed every commit and rated routine signing positively, many faced friction during setup, multi-device configuration, and repository verification. Despite signing all semester, they struggled to spot anomalous commits during verification, with over a quarter finding none. Additionally, nearly half expressed at least one misconception regarding signing guarantees or key management. Without isolating whether these difficulties stemmed from tooling, education, or understanding, we conclude that making signing easier is not enough to ensure effective security use. Rather, secure adoption also requires tools and education that support signature verification against authorized identities, interpretation of missing signatures and unknown keys, and correct reasoning about key-lifecycle operations.
Read the full abstract
Quantum Physics
Anomalous Frequency Shift in Low-Loss Superconducting Granular Aluminum Resonators
Superconducting high-kinetic inductance materials like granular aluminum (grAl) are a versatile part of the circuit quantum electrodynamics (cQED) toolbox, provided their losses at microwave frequencies are low enough. To further advance the use of grAl in quantum devices, it is indispensable to identify the dominant loss mechanisms. The standard approach pairs electromagnetic simulations of resonator geometry with measured temperature and power dependence of resonance frequency and loss rate. Most materials follow the phenomenological theory of two-level systems (TLSs) at low temperatures, resulting in an initial decrease of the resonance frequency with increasing temperature. In our work, we observe an opposite behavior at the lowest temperatures: The resonance frequency initially increases with both temperature and readout power, contradicting the standard TLS model predictions. We are able to explain a part of the data by an alternative mechanism associated with the effect of superconducting quasiparticles in a granular system with spatially-nonuniform gap. Yet, the observed change in the resonance frequency with temperature is not matched by a proportional change in the loss rate. We also observe anomalous positive frequency responses following high-energy events, characterized by several time scales. While anomalous behavior has been reported previously in grAl, the disagreement with standard models is particularly visible in our devices thanks to their exceptionally low loss rates.
Read the full abstract
Computer Vision
Elastic Token Compression for Pixel-Space Diffusion Transformers
Natural images concentrate their detail in a small fraction of the frame, yet diffusion models spend a full token on every patch, in every layer and at every timestep. The waste is largest in pixel-space models, with no autoencoder to absorb low-level redundancy first. Probing a pretrained pixel text-to-image transformer, we find its middle-block tokens redundant wherever the image is flat. The redundancy occupies connected, content-shaped regions, and exploiting it requires tokens with the same geometry. Cutting a Hilbert ordering of the patches provides them. Consecutive positions are always image neighbours, so any contiguous run is a connected region whose size and shape follow the content, and grouping in two dimensions becomes a cut in one. Existing reductions each lose part of this. Similarity merging scatters its groups, latent bottlenecks discard position, and skipping deletes what it should summarize. We cut where the model's features change most and pool each run into one region token. Our Region Token Interface (\method{}) adapts a diffusion model to these tokens, with the region count drawn at random during fine-tuning so one checkpoint serves every budget. \method{} leads prior reduction methods at matched budgets, matches dense quality at $2.0\times$ the speed, and stays close at $2.6\times$. The code and models are open-sourced at https://eduardzamfir.github.io/rti
Read the full abstract
Computer Vision
RAGDiffusion++: From Macro-Retrieval to Micro-Fidelity Alignment for Garment Generation
Standard clothing asset generation—restoring forward-facing flat-lay garment images from diverse real-world contexts—holds immense commercial value yet demands both macroscopic topological accuracy and microscopic physical fidelity. Although our previous work RAGDiffusion effectively eradicated large-scale structural hallucinations via retrieval-augmented macro-constraints, achieving industrial-grade micro-texture realism remains an unsolved bottleneck. We formally identify this limitation as High-Frequency Trajectory Collapse: supervised fine-tuning (SFT) converges to the conditional mean of the training distribution, which is dominated by smooth, low-frequency textures, causing high-frequency patterns (e.g., fabric weaves, intricate logos) to become nearly un-sampleable. Naively applying Reinforcement Learning (RL) post-training further triggers Artifact Hacking, where models exploit semantic biases in generic reward models by generating deceptive checkerboard noise. Our key insight is that RL can fundamentally reshape the sampling distribution of flow models—elevating the probability of high-fidelity trajectories under accurate reward guidance—while adversarial regularization prevents exploitation of reward blind spots. Realizing this principle requires three prerequisites: (i)inherent capacity, established through a 27,725-pair high-complexity garment dataset (STGarment-Plus) and a Dual-Image-Stream FLUX architecture upgrade; (ii)perceptive reward, provided by a novel attribute-aware reward model (Garment-RM) trained on 500K images via fine-grained contrastive learning, achieving 84.67% human preference accuracy; and (iii)hacking prevention, enforced by our Adversarial-Regularized GRPO (AR-GRPO) strategy that integrates a dynamic discriminator into the RL sampling trajectory to penalize artifacts while enriching authentic high-frequency details.
Read the full abstract
Artificial Intelligence
Understanding Deep Learning via Entropy Space Theory
Deep learning is often criticized for its theoretical research lagging behind practice. To make deep learning easier to understand, the entropy space theory is first introduced here. The entropy space can cover all the possibilities of any deep learning model by topological structure. It is independent of network parameters. Through the designed fundamental operations and norm, entropy space is proven to be a normed space within the formal axiomatic framework. Based on the theory, a unified coordinate system is proposed. It can coordinatize every state of a model and rank them by compression of the maximal value of information entropy. The theory offers a novel priori framework for mathematical fundamentals of deep learning.
Read the full abstract
Computation and Language
Modality Fault Lines: Structural Corruptions Reveal Fragile Omni-Modal Reasoning
Omni-modal large language models are increasingly evaluated on clean text–vision–audio inputs, where every channel is present, synchronized, and readily interpretable. Such scores are often taken as evidence of robust cross-modal fusion, but clean evaluation cannot tell whether success depends on stable cross-modal structure or on cues sufficient only in intact inputs. To address this gap, we define a modality fault line: a boundary at which model behavior becomes unstable when a modality remains present and human-interpretable, but its internal evidence structure is perturbed. We introduce SCEval (Structure-Corruption Evaluation) a diagnostic evaluation protocol that keeps the question, answer space, and modality channels fixed while applying controlled structural corruptions to text, vision, and audio individually and jointly. Built from $273$ human-verified tri-modal examples from Social-IQ, OmniBench, and VALOR, SCEval evaluates $15$ proprietary and open-source omni-modal systems. The results show that structural corruption lowers clean accuracy, text–vision damage forms the most stable shared fault line, and multi-modal degradation is non-additive rather than a simple function of the number of corrupted modalities. Clean omni-modal accuracy therefore does not establish that a model will remain reliable when cross-modal evidence becomes structurally unreliable.
Read the full abstract
General Relativity and Quantum Cosmology
Energy extraction processes and thermodynamics in rotating Weyl conformal black holes
This work investigates the original Penrose process (PP), as well as black hole (BH) thermodynamics, within the framework of conformal Weyl gravity. The theory is characterized by the dimensionful constants $\gamma$, which encodes the asymptotic rotational curvature of galactic rotation curves in an otherwise asymptotically flat spacetime, and $\kappa$, which is associated with the cosmological background constant $\Lambda$. We determine the locations of the event horizon and the extreme configurations for rotating BHs in this theory, and we provide both perturbative and exact expressions for the event horizon radius. The geodesic equations are restricted to the equatorial plane, and the corresponding dynamics, including energy extraction processes, are studied exclusively in this plane. As the principal results, we examine the feasibility and efficiency of the original PP, analyze BH deceleration mechanisms, and test the validity of the four laws of BH thermodynamics in the context of conformal Weyl gravity.
Read the full abstract
Strongly Correlated Electrons
Crystal electric field excitations and an effective-spin-$1/2$ ground doublet in the hyperkagome magnet Yb$_3$Sc$_2$Ga$_3$O$_{12}$
We use inelastic neutron scattering (INS) to determine the crystal electric field (CEF) excitations of Yb$^{3+}$ in the rare-earth hyperkagome magnet Yb$_3$Sc$_2$Ga$_3$O$_{12}$. Three nearly dispersionless magnetic excitations are observed near 58, 68, and 74~meV, corresponding to transitions from the ground-state Kramers doublet to the three excited doublets of the $J=7/2$ multiplet. A Stevens-operator analysis constrained by the local $222$ ($D_2$) symmetry reproduces the excitation energies and spectral weights and yields an Ising-type ground-state $g$ tensor. The first excited doublet lies approximately 58~meV above the ground state, establishing a well-isolated effective $J_{\mathrm{eff}}=1/2$ degree of freedom over the low-temperature regime relevant to collective magnetism. Notably, the directly measured CEF spectrum substantially revises the level scheme previously inferred from bulk measurements, while preserving the essential low-energy pseudospin description. Two independent fitting protocols give consistent excitation energies, ground-doublet wave functions, and $g$ tensors, despite the nonuniqueness of the individual CEF parameters. The resulting single-ion model also reproduces the characteristic susceptibility, magnetization, and field evolution of the Schottky anomaly in specific heat. These results establish the microscopic single-ion basis needed to construct an effective exchange Hamiltonian and to interpret future measurements of low-energy collective excitations in Yb$_3$Sc$_2$Ga$_3$O$_{12}$.
Read the full abstract
High Energy Physics (Phenomenology)
Probing Ultra-Compressed Scotogenic Dark Matter at the HL-LHC via 4D Spacetime Tracking
The ultra-compressed scotogenic regime, in which a 2\,GeV mass splitting between the charged inert scalar and the dark-matter fermion is motivated by and compatible with co-annihilation to the observed relic density for suitable choices of the remaining model parameters, produces decay leptons for which existing LHC displaced-lepton and disappearing-track searches retain only partial, unoptimised acceptance, leaving room for a dedicated timing-assisted strategy. We propose exploiting the 30\,ps timing resolution of the HL-LHC precision timing detectors as a fourth observable, converting the macroscopic scalar decay length of 100–1000\,mm into an arrival-time delay of 70–700\,ps—more than three orders of magnitude above the sub-picosecond delays of all prompt Standard Model backgrounds. Operating at the level of a low-$p_T$ disappearing-track stub matched to a delayed timing hit, without requiring full lepton reconstruction, our 4D timing strategy achieves signal efficiencies of 12–36% across the probed mass range (14% at the 200\,GeV benchmark) and zero background across $4.20\times10^6$ simulated prompt SM events, projecting 95%\,C.L.\ exclusion up to scalar masses of $\simeq\!670$\,GeV at 3000\,fb$^{-1}$ under conservative background assumptions. The scalar lifetime is set by the Yukawa coupling $y$ through $c\tau\propto y^{-2}$, while sub-eV neutrino masses constrain only the product $y^2\lambda_5$; the small couplings that place the charged scalar in the optimal timing window are therefore compatible with, though not uniquely fixed by, the radiative neutrino-mass mechanism, making the co-annihilation corridor—conventionally the most elusive regime—an accessible target for 4D spacetime tracking at the HL-LHC.
Read the full abstract
Data Structures and Algorithms
A Simplified Analysis of the Good-Bad $3/2$-Approximation Algorithm for Some Minimum-Cost Graph Problems
In this paper, we consider an easy greedy approximation algorithm, the good-bad algorithm, introduced by Couëtoux for finding a minimum-cost set of edges such that every connected component has at least $k$ vertices. Couëtoux proves that the good-bad algorithm achieves a $3/2$-approximation for this problem. Davis and Williamson extend this result to the more general problem of finding a minimum-cost edge set that contains at least one edge from every cut $S\subseteq V$ satisfying $h(S) = 1$ where $h:2^V \rightarrow \{0,1\}$ is downward monotone; that is, $h(S) = 1$ implies $h(T) = 1$ for every nonempty subset $T \subseteq S$. The original problem corresponds to $h(S) =1$ when $|S|<k$. We give a simplified analysis of the good-bad algorithm for downward monotone functions.
Read the full abstract
Optimization and Control
Value of Information in Dynamic Decision Making
We study the value of information in predicting the evolving state of a finite Markov chain. At each stage, a decision maker chooses a state and observes only whether the current state of the chain matches her choice; the resulting information is used to make a prediction on the state at the final stage of the problem. We show that, when the chain starts from an invariant distribution, the optimal terminal value is non-decreasing with the number of observations and converges at a uniform exponential rate. We introduce the predictive learning index, which measures whether all attainable informational value is extracted after finitely many observations, and show that both finite and infinite indices may occur. In contrast, for nonstationary initial distributions, the value may strictly decrease with the horizon.
Read the full abstract
Systems and Control
Stochastic Nonlinear Model Predictive Control with Gaussian Mixture Uncertainty Propagation
We propose a novel Stochastic Nonlinear Model Predictive Control (SNMPC) framework for nonlinear systems with additive noise. Building on recent advances in nonlinear uncertainty propagation, we show that the state distribution of the system can be tractably approximated over time by Gaussian mixture distributions, with formal error bounds in Wasserstein distance. This representation yields closed-form expressions for expected costs and chance constraints, which become exact for affine constraints and exact up to a constant for quadratic costs. Consequently, the resulting control problem can be solved efficiently via nonlinear programming, while providing formal open-loop guarantees of correctness and asymptotic optimality. Experiments on a set of benchmarks demonstrate that the proposed approach compares favorably with existing methods in nonlinear settings with multi-modal disturbances, where standard approaches lead to poorly scaled solutions and unsafe or overly conservative control actions.