# Research Papers — 2026-03 · bundle page 54 of 243

> Full-text markdown of papers #2651–2700 of 12123 published in 2026-03, newest first. AI-readable mirrors of the full 12123-paper 2026-03 archive (50 papers per file).

Canonical page: https://aipod.dev/papers/archive/2026-03/page/54
[← Previous bundle](https://aipod.dev/papers/archive/2026-03/page/53.md)
[Next bundle →](https://aipod.dev/papers/archive/2026-03/page/55.md)

### LLMs Do Not Grade Essays Like Humans

Authors: Jerin George Mathew, Sumayya Taher, Anindita Kundu, Denilson Barbosa
Published: 2026-03-24
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2603.23714v1

Large language models have recently been proposed as tools for automated essay scoring, but their agreement with human grading remains unclear. In this work, we evaluate how LLM-generated scores compare with human grades and analyze the grading behavior of several models from the GPT and Llama families in an out-of-the-box setting, without task-specific training. Our results show that agreement between LLM and human scores remains relatively weak and varies with essay characteristics. In particular, compared to human raters, LLMs tend to assign higher scores to short or underdeveloped essays, while assigning lower scores to longer essays that contain minor grammatical or spelling errors. We also find that the scores generated by LLMs are generally consistent with the feedback they generate: essays receiving more praise tend to receive higher scores, while essays receiving more criticism tend to receive lower scores. These results suggest that LLM-generated scores and feedback follow coherent patterns but rely on signals that differ from those used by human raters, resulting in limited alignment with human grading practices. Nevertheless, our work shows that LLMs produce feedback that is consistent with their grading and that they can be reliably used in supporting essay scoring.

---

### Mind the Hitch: Dynamic Calibration and Articulated Perception for Autonomous Trucks

Authors: Morui Zhu, Yongqi Zhu, Song Fu, Qing Yang
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23711v2

Autonomous trucking poses unique challenges due to articulated tractor-trailer geometry, and time-varying sensor poses caused by the fifth-wheel joint and trailer flex. Existing perception and calibration methods assume static baselines or rely on high-parallax and texture-rich scenes, limiting their reliability under real-world settings. We propose dCAP (dynamic Calibration and Articulated Perception), a vision-based framework that continuously estimates the 6-DoF (degree of freedom) relative pose between tractor and trailer cameras. dCAP employs a transformer with cross-view and temporal attention to robustly aggregate spatial cues while maintaining temporal consistency, enabling accurate perception under rapid articulation and occlusion. Integrated with BEVFormer, dCAP improves 3D object detection by replacing static calibration with dynamically predicted extrinsics. To facilitate evaluation, we introduce STT4AT, a CARLA-based benchmark simulating semi-trailer trucks with synchronized multi-sensor suites and time-varying inter-rig geometry across diverse environments. Experiments demonstrate that dCAP achieves stable, accurate perception while addressing the limitations of static calibration in autonomous trucking. The dataset, development kit, and source code will be publicly released.

---

### An In-Depth Study of Filter-Agnostic Vector Search on a PostgreSQL Database System: (Experiments and Analysis)

Authors: Duo Lu, Helena Caminal, Manos Chatzakis, Yannis Papakonstantinou, Yannis Chronis, Vaibhav Jain, Fatma Özcan
Published: 2026-03-24
Categories: cs.DB, cs.AI, cs.IR
arXiv: https://arxiv.org/abs/2603.23710v1

Filtered Vector Search (FVS) is critical for supporting semantic search and GenAI applications in modern database systems. However, existing research most often evaluates algorithms in specialized libraries, making optimistic assumptions that do not align with enterprise-grade database systems. Our work challenges this premise by demonstrating that in a production-grade database system, commonly made assumptions do not hold, leading to performance characteristics and algorithmic trade-offs that are fundamentally different from those observed in isolated library settings. This paper presents the first in-depth analysis of filter-agnostic FVS algorithms within a production PostgreSQL-compatible system. We systematically evaluate post-filtering and inline-filtering strategies across a wide range of selectivities and correlations. Our central finding is that the optimal algorithm is not dictated by the cost of distance computations alone, but that system-level overheads that come from both distance computations and filter operations (like page accesses and data retrieval) play a significant role. We demonstrate that graph-based approaches (such as NaviX/ACORN) can incur prohibitive numbers of filter checks and system-level overheads, compared with clustering-based indexes such as ScaNN, often canceling out their theoretical benefits in real-world database environments. Ultimately, our findings provide the database community with crucial insights and practical guidelines, demonstrating that the optimal choice for a filter-agnostic FVS algorithm is not absolute, but rather a system-aware decision contingent on the interplay between workload characteristics and the underlying costs of data access in a real-world database architecture.

---

### The Diminishing Returns of Early-Exit Decoding in Modern LLMs

Authors: Rui Wei, Rui Du, Hanfei Yu, Devesh Tiwari, Jian Li, Zhaozhuo Xu, Hao Wang
Published: 2026-03-24
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2603.23701v1

In Large Language Model (LLM) inference, early-exit refers to stopping computation at an intermediate layer once the prediction is sufficiently confident, thereby reducing latency and cost. However, recent LLMs adopt improved pretraining recipes and architectures that reduce layer redundancy, potentially limiting early-exit opportunities. We re-evaluate layer-wise early-exit in modern LLMs and analyze how intermediate representations evolve during training. We introduce a metric to quantify a model's intrinsic suitability for early-exit and propose a benchmark for researchers to explore the potential early-exit benefits on different models and workloads. Our results show a diminishing trend in early-exit effectiveness across newer model generations. We further find that dense transformers generally offer greater early-exit potential than Mixture-of-Experts and State Space Models. In addition, larger models, particularly those with more than 20 billion parameters, and base pretrained models without specialized tuning tend to exhibit higher early-exit potential.

---

### CoRe: Joint Optimization with Contrastive Learning for Medical Image Registration

Authors: Eytan Kats, Christoph Grossbroehmer, Ziad Al-Haj Hemidi, Fenja Falta, Wiebke Heyer, Mattias P. Heinrich
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23694v2

Medical image registration is a fundamental task in medical image analysis, enabling the alignment of images from different modalities or time points. However, intensity inconsistencies and nonlinear tissue deformations pose significant challenges to the robustness of registration methods. Recent approaches leveraging self-supervised representation learning show promise by pre-training feature extractors to generate robust anatomical embeddings, that farther used for the registration. In this work, we propose a novel framework that integrates equivariant contrastive learning directly into the registration model. Our approach leverages the power of contrastive learning to learn robust feature representations that are invariant to tissue deformations. By jointly optimizing the contrastive and registration objectives, we ensure that the learned representations are not only informative but also suitable for the registration task. We evaluate our method on abdominal and thoracic image registration tasks, including both intra-patient and inter-patient scenarios. Experimental results demonstrate that the integration of contrastive learning directly into the registration framework significantly improves performance, surpassing strong baseline methods.

---

### Adaptive Gaussian Process Search for Simulation-Based Sample Size Estimation in Clinical Prediction Models: Validation of the pmsims R Package

Authors: Oyebayo Ridwan Olaniran, Diana Shamsutdinova, Sarah Markham, Felix Zimmer, Daniel Stahl, Gordon Forbes, Ewan Carr
Published: 2026-03-24
Categories: stat.CO, stat.AP, stat.ME, stat.ML
arXiv: https://arxiv.org/abs/2603.23688v1

Background: Determining an adequate sample size is essential for developing reliable and generalisable clinical prediction models, yet practical guidance on selecting appropriate methods remains limited. Existing analytical and simulation-based approaches often rely on restrictive assumptions and focus on mean-based criteria. We present and validate pmsims, an R package that uses Gaussian process surrogate modelling to provide a flexible and computationally efficient simulation-based framework for sample size determination across diverse prediction settings. Methods: We conducted a comprehensive simulation study with two aims. First, we compared three search engines implemented in pmsims: a Gaussian process-based adaptive method, a deterministic bisection method, and a hybrid approach, across binary, continuous, and survival outcomes. Second, we benchmarked the best-performing pmsims engine against existing analytical (pmsampsize) and simulation-based (samplesizedev) methods, evaluating recommended sample sizes, computational time, and achieved performance on large independent validation datasets. Results: The Gaussian process-based method consistently produced the most stable sample size estimates, particularly in low-signal, high-dimensional settings. In benchmarking, pmsims achieved performance close to prespecified targets across all outcome types, matching simulation-based approaches and outperforming analytical methods in more challenging scenarios. Conclusions: pmsims provides an efficient and flexible framework for principled sample size planning in clinical prediction modelling, requiring fewer model evaluations than non-adaptive simulation approaches.

---

### AdvSplat: Adversarial Attacks on Feed-Forward Gaussian Splatting Models

Authors: Yiran Qiao, Yiren Lu, Yunlai Zhou, Rui Yang, Linlin Hou, Yu Yin, Jing Ma
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23686v1

3D Gaussian Splatting (3DGS) is increasingly recognized as a powerful paradigm for real-time, high-fidelity 3D reconstruction. However, its per-scene optimization pipeline limits scalability and generalization, and prevents efficient inference. Recently emerged feed-forward 3DGS models address these limitations by enabling fast reconstruction from a few input views after large-scale pretraining, without scene-specific optimization. Despite their advantages and strong potential for commercial deployment, the use of neural networks as the backbone also amplifies the risk of adversarial manipulation. In this paper, we introduce AdvSplat, the first systematic study of adversarial attacks on feed-forward 3DGS. We first employ white-box attacks to reveal fundamental vulnerabilities of this model family. We then develop two improved, practically relevant, query-efficient black-box algorithms that optimize pixel-space perturbations via a frequency-domain parameterization: one based on gradient estimation and the other gradient-free, without requiring any access to model internals. Extensive experiments across multiple datasets demonstrate that AdvSplat can significantly disrupt reconstruction results by injecting imperceptible perturbations into the input images. Our findings surface an overlooked yet urgent problem in this domain, and we hope to draw the community's attention to this emerging security and robustness challenge.

---

### The Economics of Builder Saturation in Digital Markets

Authors: Armin Catovic
Published: 2026-03-24
Categories: econ.TH, cs.CY, cs.GT, cs.LG, econ.GN
arXiv: https://arxiv.org/abs/2603.23685v2

Recent advances in generative AI systems have dramatically reduced the cost of digital production, fueling narratives that widespread participation in software creation will yield a proliferation of viable companies. This paper challenges that assumption. We introduce the Builder Saturation Effect, formalizing a model in which production scales elastically but human attention remains finite. In markets with near-zero marginal costs and free entry, increases in the number of producers dilute average attention and returns per producer, even as total output expands. Extending the framework to incorporate quality heterogeneity and reinforcement dynamics, we show that equilibrium outcomes exhibit declining average payoffs and increasing concentration, consistent with power-law-like distributions. These results suggest that AI-enabled, democratised production is more likely to intensify competition and produce winner-take-most outcomes than to generate broadly distributed entrepreneurial success. Contribution type: This paper is primarily a work of synthesis and applied formalisation. The individual theoretical ingredients - attention scarcity, free-entry dilution, superstar effects, preferential attachment - are well established in their respective literatures. The contribution is to combine them into a unified framework and direct the resulting predictions at a specific contemporary claim about AI-enabled entrepreneurship.

---

### MoCHA: Denoising Caption Supervision for Motion-Text Retrieval

Authors: Nikolai Warner, Cameron Ethan Taylor, Irfan Essa, Apaar Sadhwani
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23684v2

Text-motion retrieval systems learn shared embedding spaces from motion-caption pairs via contrastive objectives. However, each caption is not a deterministic label but a sample from a distribution of valid descriptions: different annotators produce different text for the same motion, mixing motion-recoverable semantics (action type, body parts, directionality) with annotator-specific style and inferred context that cannot be determined from 3D joint coordinates alone. Standard contrastive training treats each caption as the single positive target, overlooking this distributional structure and inducing within-motion embedding variance that weakens alignment. We propose MoCHA, a text canonicalization framework that reduces this variance by projecting each caption onto its motion-recoverable content prior to encoding, producing tighter positive clusters and better-separated embeddings. Canonicalization is a general principle: even deterministic rule-based methods improve cross-dataset transfer, though learned canonicalizers provide substantially larger gains. We present two learned variants: an LLM-based approach (GPT-5.2) and a distilled FlanT5 model requiring no LLM at inference time. MoCHA operates as a preprocessing step compatible with any retrieval architecture. Applied to MoPa (MotionPatches), MoCHA sets a new state of the art on both HumanML3D (H) and KIT-ML (K): the LLM variant achieves 13.9% T2M R@1 on H (+3.1pp) and 24.3% on K (+10.3pp), while the LLM-free T5 variant achieves gains of +2.5pp and +8.1pp. Canonicalization reduces within-motion text-embedding variance by 11-19% and improves cross-dataset transfer substantially, with H to K improving by 94% and K to H by 52%, demonstrating that standardizing the language space yields more transferable motion-language representations.

---

### Assessment Design in the AI Era: A Method for Identifying Items Functioning Differentially for Humans and Chatbots

Authors: Licol Zeinfeld, Alona Strugatski, Ziva Bar-Dov, Ron Blonder, Shelley Rap, Giora Alexandron
Published: 2026-03-24
Categories: cs.HC, cs.AI
arXiv: https://arxiv.org/abs/2603.23682v2

The rapid adoption of large language models (LLMs) in education raises profound challenges for assessment design. To adapt assessments to the presence of LLM-based tools, it is crucial to characterize the strengths and weaknesses of LLMs in a generalizable, valid and reliable manner. However, current LLM evaluations often rely on descriptive statistics derived from benchmarks, and little research applies theory-grounded measurement methods to characterize LLM capabilities relative to human learners in ways that directly support assessment design. Here, by combining educational data mining and psychometric theory, we introduce a statistically principled approach for identifying items on which humans and LLMs show systematic response differences, pinpointing where assessments may be most vulnerable to AI misuse, and which task dimensions make problems particularly easy or difficult for generative AI. The method is based on Differential Item Functioning (DIF) analysis -- traditionally used to detect bias across demographic groups -- together with negative control analysis and item-total correlation discrimination analysis. It is evaluated on responses from human learners and six leading chatbots (ChatGPT-4o \& 5.2, Gemini 1.5 \& 3 Pro, Claude 3.5 \& 4.5 Sonnet) to two instruments: a high school chemistry diagnostic test and a university entrance exam. Subject-matter experts then analyzed DIF-flagged items to characterize task dimensions associated with chatbot over- or under-performance. Results show that DIF-informed analytics provide a robust framework for understanding where LLM and human capabilities diverge, and highlight their value for improving the design of valid, reliable, and fair assessment in the AI era.

---

### Learning What Can Be Picked: Active Reachability Estimation for Efficient Robotic Fruit Harvesting

Authors: Nur Afsa Syeda, Mohamed Elmahallawy, Luis Fernando de la Torre, John Miller
Published: 2026-03-24
Categories: cs.RO, cs.AI
arXiv: https://arxiv.org/abs/2603.23679v2

Agriculture remains a cornerstone of global health and economic sustainability, yet labor-intensive tasks such as harvesting high-value crops continue to face growing workforce shortages. Robotic harvesting systems offer a promising solution; however, their deployment in unstructured orchard environments is constrained by inefficient perception-to-action pipelines. In particular, existing approaches often rely on exhaustive inverse kinematics or motion planning to determine whether a target fruit is reachable, leading to unnecessary computation and delayed decision-making. Our approach combines RGB-D perception with active learning to directly learn reachability as a binary decision problem. We then leverage active learning to selectively query the most informative samples for reachability labeling, significantly reducing annotation effort while maintaining high predictive accuracy. Extensive experiments demonstrate that the proposed framework achieves accurate reachability prediction with substantially fewer labeled samples, yielding approximately 6--8% higher accuracy than random sampling and enabling label-efficient adaptation to new orchard configurations. Among the evaluated strategies, entropy- and margin-based sampling outperform Query-by-Committee and standard uncertainty sampling in low-label regimes, while all strategies converge to comparable performance as the labeled set grows. These results highlight the effectiveness of active learning for task-level perception in agricultural robotics and position our approach as a scalable alternative to computation-heavy kinematic reachability analysis. Our code is available through https://github.com/wsu-cyber-security-lab-ai/active-learning.

---

### PLACID: Privacy-preserving Large language models for Acronym Clinical Inference and Disambiguation

Authors: Manjushree B. Aithal, Ph. D., Alexander Kotz, James Mitchell, Ph. D
Published: 2026-03-24
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2603.23678v1

Large Language Models (LLMs) offer transformative solutions across many domains, but healthcare integration is hindered by strict data privacy constraints. Clinical narratives are dense with ambiguous acronyms, misinterpretation these abbreviations can precipitate severe outcomes like life-threatening medication errors. While cloud-dependent LLMs excel at Acronym Disambiguation, transmitting Protected Health Information to external servers violates privacy frameworks. To bridge this gap, this study pioneers the evaluation of small-parameter models deployed entirely on-device to ensure privacy preservation. We introduce a privacy-preserving cascaded pipeline leveraging general-purpose local models to detect clinical acronyms, routing them to domain-specific biomedical models for context-relevant expansions. Results reveal that while general instruction-following models achieve high detection accuracy (~0.988), their expansion capabilities plummet (~0.655). Our cascaded approach utilizes domain-specific medical models to increase expansion accuracy to (~0.81). This novel work demonstrates that privacy-preserving, on-device (2B-10B) models deliver high-fidelity clinical acronym disambiguation support.

---

### Prototype Fusion: A Training-Free Multi-Layer Approach to OOD Detection

Authors: Shreen Gul, Mohamed Elmahallawy, Ardhendu Tripathy, Sanjay Madria
Published: 2026-03-24
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2603.23677v2

Deep learning models are increasingly deployed in safety-critical applications, where reliable out-of-distribution (OOD) detection is essential to ensure robustness. Existing methods predominantly rely on the penultimate-layer activations of neural networks, assuming they encapsulate the most informative in-distribution (ID) representations. In this work, we revisit this assumption to show that intermediate layers encode equally rich and discriminative information for OOD detection. Based on this observation, we propose a simple yet effective model-agnostic approach that leverages internal representations across multiple layers. Our scheme aggregates features from successive convolutional blocks, computes class-wise mean embeddings, and applies L_2 normalization to form compact ID prototypes capturing class semantics. During inference, cosine similarity between test features and these prototypes serves as an OOD score--ID samples exhibit strong affinity to at least one prototype, whereas OOD samples remain uniformly distant. Extensive experiments on state-of-the-art OOD benchmarks across diverse architectures demonstrate that our approach delivers robust, architecture-agnostic performance and strong generalization for image classification. Notably, it improves AUROC by up to 4.41% and reduces FPR by 13.58%, highlighting multi-layer feature aggregation as a powerful yet underexplored signal for OOD detection, challenging the dominance of penultimate-layer-based methods. Our code is available at: https://github.com/sgchr273/cosine-layers.git.

---

### Grounding Vision and Language to 3D Masks for Long-Horizon Box Rearrangement

Authors: Ashish Malik, Caleb Lowe, Aayam Shrestha, Stefan Lee, Fuxin Li, Alan Fern
Published: 2026-03-24
Categories: cs.AI, cs.RO
arXiv: https://arxiv.org/abs/2603.23676v1

We study long-horizon planning in 3D environments from under-specified natural-language goals using only visual observations, focusing on multi-step 3D box rearrangement tasks. Existing approaches typically rely on symbolic planners with brittle relational grounding of states and goals, or on direct action-sequence generation from 2D vision-language models (VLMs). Both approaches struggle with reasoning over many objects, rich 3D geometry, and implicit semantic constraints. Recent advances in 3D VLMs demonstrate strong grounding of natural-language referents to 3D segmentation masks, suggesting the potential for more general planning capabilities. We extend existing 3D grounding models and propose Reactive Action Mask Planner (RAMP-3D), which formulates long-horizon planning as sequential reactive prediction of paired 3D masks: a "which-object" mask indicating what to pick and a "which-target-region" mask specifying where to place it. The resulting system processes RGB-D observations and natural-language task specifications to reactively generate multi-step pick-and-place actions for 3D box rearrangement. We conduct experiments across 11 task variants in warehouse-style environments with 1-30 boxes and diverse natural-language constraints. RAMP-3D achieves 79.5% success rate on long-horizon rearrangement tasks and significantly outperforms 2D VLM-based baselines, establishing mask-based reactive policies as a promising alternative to symbolic pipelines for long-horizon planning.

---

### Bio-Inspired Event-Based Visual Servoing for Ground Robots

Authors: Maral Mordad, Kian Behzad, Debojyoti Biswas, Noah J. Cowan, Milad Siami
Published: 2026-03-24
Categories: cs.RO, cs.CV
arXiv: https://arxiv.org/abs/2603.23672v2

Biological sensory systems are inherently adaptive, filtering out constant stimuli and prioritizing relative changes, likely enhancing computational and metabolic efficiency. Inspired by active sensing behaviors across a wide range of animals, this paper introduces a principled 1D event-based visual servoing framework for ground robots operating in structured environments. Utilizing a Dynamic Vision Sensor (DVS), we demonstrate that by applying a fixed spatial kernel to the asynchronous event stream generated from structured logarithmic intensity-change patterns, the resulting net event flux analytically isolates specific combinations of kinematic states. We establish a generalized theoretical bound for this event rate estimator and show that linear and quadratic spatial profiles isolate the robot's velocity and position-velocity product, respectively. Leveraging these properties, we employ a multi-pattern stimulus to directly synthesize a nonlinear state feedback term entirely without traditional state estimation. To overcome the inescapable loss of linear observability at equilibrium inherent in event sensing, we propose a bio-inspired active sensing limit-cycle controller. Experimental validation on a 1/10-scale autonomous ground vehicle confirms the efficacy, extreme low-latency, and computational efficiency of the proposed direct-sensing approach.

---

### Estimating Individual Tree Height and Species from UAV Imagery

Authors: Jannik Endres, Etienne Laliberté, David Rolnick, Arthur Ouaknine
Published: 2026-03-24
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2603.23669v2

Accurate estimation of forest biomass, a major carbon sink, relies heavily on tree-level traits such as height and species. Unoccupied Aerial Vehicles (UAVs) capturing high-resolution imagery from a single RGB camera offer a cost-effective and scalable approach for mapping and measuring individual trees. We introduce BIRCH-Trees, the first benchmark for individual tree height and species estimation from tree-centered UAV images, spanning three datasets: temperate forests, tropical forests, and boreal plantations. We also present DINOvTree, a unified approach using a Vision Foundation Model (VFM) backbone with task-specific heads for simultaneous height and species prediction. Through extensive evaluations on BIRCH-Trees, we compare DINOvTree against commonly used vision methods, including VFMs, as well as biological allometric equations. We find that DINOvTree achieves top overall results with accurate height predictions and competitive classification accuracy while using only 54% to 58% of the parameters of the second-best approach.

---

### Energy Efficient Software Hardware CoDesign for Machine Learning: From TinyML to Large Language Models

Authors: Mohammad Saleh Vahdatpour, Yanqing Zhang
Published: 2026-03-24
Categories: cs.AR, cs.LG
arXiv: https://arxiv.org/abs/2603.23668v1

The rapid deployment of machine learning across platforms from milliwatt-class TinyML devices to large language models has made energy efficiency a primary constraint for sustainable AI. Across these scales, performance and energy are increasingly limited by data movement and memory-system behavior rather than by arithmetic throughput alone. This work reviews energy efficient software hardware codesign methods spanning edge inference and training to datacenter-scale LLM serving, covering accelerator architectures (e.g., ASIC/FPGA dataflows, processing-/compute-in-memory designs) and system-level techniques (e.g., partitioning, quantization, scheduling, and runtime adaptation). We distill common design levers and trade-offs, and highlight recurring gaps including limited cross-platform generalization, large and costly co-design search spaces, and inconsistent benchmarking across workloads and deployment settings. Finally, we outline a hierarchical decomposition perspective that maps optimization strategies to computational roles and supports incremental adaptation, offering practical guidance for building energy and carbon aware ML systems.

---

### Echoes: A semantically-aligned music deepfake detection dataset

Authors: Octavian Pascu, Dan Oneata, Horia Cucu, Nicolas M. Muller
Published: 2026-03-24
Categories: cs.SD, cs.AI, eess.AS
arXiv: https://arxiv.org/abs/2603.23667v3

We introduce Echoes, a new dataset for music deepfake detection designed for training and benchmarking detectors under realistic and provider-diverse conditions. Echoes comprises 4,468 tracks (131 hours of audio) spanning multiple genres (pop, rock, electronic), and includes content generated by ten popular AI music generation systems. To prevent shortcut learning and promote robust generalization, the dataset is deliberately constructed to be challenging, enforcing semantic-level alignment between spoofed audio and bona fide references. This alignment is achieved by conditioning generated audio samples directly on bona-fide waveforms or song descriptors. We evaluate Echoes in a cross-dataset setting against three existing AI-generated music datasets using state-of-the-art Wav2Vec2 XLS-R 2B representations. Results show that (i) Echoes is the hardest in-domain dataset; (ii) detectors trained on existing datasets transfer poorly to Echoes; (iii) training on Echoes yields the strongest generalization performance. These findings suggest that provider diversity and semantic alignment help learn more transferable detection cues.

---

### GTO Wizard Benchmark

Authors: Marc-Antoine Provost, Nejc Ilenic, Christopher Solinas, Philippe Beardsell
Published: 2026-03-24
Categories: cs.AI
arXiv: https://arxiv.org/abs/2603.23660v1

We introduce GTO Wizard Benchmark, a public API and standardized evaluation framework for benchmarking algorithms in Heads-Up No-Limit Texas Hold'em (HUNL). The benchmark evaluates agents against GTO Wizard AI, a state-of-the-art superhuman poker agent that approximates Nash Equilibria, and defeated Slumbot, the 2018 Annual Computer Poker Competition champion and previous strongest publicly accessible HUNL benchmark, by $19.4$ $\pm$ $4.1$ bb/100. Variance is a fundamental challenge in poker evaluation; we address this by integrating AIVAT, a provably unbiased variance reduction technique that achieves equivalent statistical significance with ten times fewer hands than naive Monte Carlo evaluation. We conduct a comprehensive benchmarking study of state-of-the-art large language models under zero-shot conditions, including GPT-5.4, Claude Opus 4.6, Gemini 3.1 Pro, Grok 4, and others. Initial results and analysis reveal dramatic progress in LLM reasoning over recent years, yet all models remain far below the baseline established by our benchmark. Qualitative analysis reveals clear opportunities for improvement, including representation and the ability to reason over hidden states. This benchmark provides researchers with a precise and quantifiable setting to evaluate advances in planning and reasoning in multi-agent systems with partial observability.

---

### Probing Ethical Framework Representations in Large Language Models: Structure, Entanglement, and Methodological Challenges

Authors: Weilun Xu, Alexander Rusnak, Frederic Kaplan
Published: 2026-03-24
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2603.23659v1

When large language models make ethical judgments, do their internal representations distinguish between normative frameworks, or collapse ethics into a single acceptability dimension? We probe hidden representations across five ethical frameworks (deontology, utilitarianism, virtue, justice, commonsense) in six LLMs spanning 4B--72B parameters. Our analysis reveals differentiated ethical subspaces with asymmetric transfer patterns -- e.g., deontology probes partially generalize to virtue scenarios while commonsense probes fail catastrophically on justice. Disagreement between deontological and utilitarian probes correlates with higher behavioral entropy across architectures, though this relationship may partly reflect shared sensitivity to scenario difficulty. Post-hoc validation reveals that probes partially depend on surface features of benchmark templates, motivating cautious interpretation. We discuss both the structural insights these methods provide and their epistemological limitations.

---

### Boost Like a (Var)Pro: Trust-Region Gradient Boosting via Variable Projection

Authors: Abhijit Chowdhary, Elizabeth Newman, Deepanshu Verma
Published: 2026-03-24
Categories: cs.LG, math.NA, math.OC
arXiv: https://arxiv.org/abs/2603.23658v1

Gradient boosting, a method of building additive ensembles from weak learners, has established itself as a practical and theoretically-motivated approach to approximate functions, especially using decision tree weak learners. Comparable methods for smooth parametric learners, such as neural networks, remain less developed in both training methodology and theory. To this end, we introduce \texttt{VPBoost} ({\bf V}ariable {\bf P}rojection {\bf Boost}ing), a gradient boosting algorithm for separable smooth approximators, i.e., models with a smooth nonlinear featurizer followed by a final linear mapping. \texttt{VPBoost} fuses variable projection, a training paradigm for separable models that enforces optimality of the linear weights, with a second-order weak learning strategy. The combination of second-order boosting, separable models, and variable projection give rise to a closed-form solution for the optimal linear weights and a natural interpretation of \VPBoost as a functional trust-region method. We thereby leverage trust-region theory to prove \VPBoost converges to a stationary point under mild geometric conditions and, under stronger assumptions, achieves a superlinear convergence rate. Comprehensive numerical experiments on synthetic data, image recognition, and scientific machine learning benchmarks demonstrate that \VPBoost learns an ensemble with improved evaluation metrics in comparison to gradient-descent-based boosting and attains competitive performance relative to an industry-standard decision tree boosting algorithm.

---

### Ethio-ASR: Joint Multilingual Speech Recognition and Language Identification for Ethiopian Languages

Authors: Badr M. Abdullah, Israel Abebe Azime, Atnafu Lambebo Tonja, Jesujoba O. Alabi, Abel Mulat Alemu, Eyob G. Hagos, Bontu Fufa Balcha, Mulubrhan A. Nerea, Debela Desalegn Yadeta, Dagnachew Mekonnen Marilign, Amanuel Temesgen Fentahun, Tadesse Kebede et al.
Published: 2026-03-24
Categories: cs.CL
arXiv: https://arxiv.org/abs/2603.23654v1

We present Ethio-ASR, a suite of multilingual CTC-based automatic speech recognition (ASR) models jointly trained on five Ethiopian languages: Amharic, Tigrinya, Oromo, Sidaama, and Wolaytta. These languages belong to the Semitic, Cushitic, and Omotic branches of the Afroasiatic family, and remain severely underrepresented in speech technology despite being spoken by the vast majority of Ethiopia's population. We train our models on the recently released WAXAL corpus using several pre-trained speech encoders and evaluate against strong multilingual baselines, including OmniASR. Our best model achieves an average WER of 30.48% on the WAXAL test set, outperforming the best OmniASR model with substantially fewer parameters. We further provide a comprehensive analysis of gender bias, the contribution of vowel length and consonant gemination to ASR errors, and the training dynamics of multilingual CTC models. Our models and codebase are publicly available to the research community.

---

### Foundation Model Embeddings Meet Blended Emotions: A Multimodal Fusion Approach for the BLEMORE Challenge

Authors: Masoumeh Chapariniya, Aref Farhadipour, Sarah Ebling, Volker Dellwo, Teodora Vukovic
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23650v1

We present our system for the BLEMORE Challenge at FG 2026 on blended emotion recognition with relative salience prediction. Our approach combines six encoder families through late probability fusion: an S4D-ViTMoE face encoder adapted with soft-label KL training, frozen layer-selective Wav2Vec2 audio features, finetuned body-language encoders (TimeSformer, VideoMAE), and -- for the first time in emotion recognition -- Gemini Embedding 2.0, a large multimodal model whose video embeddings produce competitive presence accuracy (ACCP = 0.320) from only 2 seconds of input. Three key findings emerge from our experiments: selecting prosody-encoding layers (6--12) from frozen Wav2Vec2 outperforms end-to-end finetuning (Score 0.207 vs. 0.161), as the non-verbal nature of BLEMORE audio makes phonetic layers irrelevant; the post-processing salience threshold $β$ varies from 0.05 to 0.43 across folds, revealing that personalized expression styles are the primary bottleneck; and task-adapted encoders collectively receive 62\% of ensemble weight over general-purpose baselines. Our 12-encoder system achieves Score = 0.279 (ACCP = 0.391, ACCS = 0.168) on the test set, placing 6th.

---

### λSplit: Self-Supervised Content-Aware Spectral Unmixing for Fluorescence Microscopy

Authors: Federico Carrara, Talley Lambert, Mehdi Seifi, Florian Jug
Published: 2026-03-24
Categories: cs.CV, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2603.23647v3

In fluorescence microscopy, spectral unmixing aims to recover individual fluorophore concentrations from spectral images that capture mixed fluorophore emissions. Since classical methods operate pixel-wise and rely on least-squares fitting, their performance degrades with increasingly overlapping emission spectra and higher levels of noise, suggesting that a data-driven approach that can learn and utilize a structural prior might lead to improved results. Learning-based approaches for spectral imaging do exist, but they are either not optimized for microscopy data or are developed for very specific cases that are not applicable to fluorescence microscopy settings. To address this, we propose λSplit, a physics-informed deep generative model that learns a conditional distribution over concentration maps using a hierarchical Variational Autoencoder. A fully differentiable Spectral Mixer enforces consistency with the image formation process, while the learned structural priors enable state-of-the-art unmixing and implicit noise removal. We demonstrate λSplit on 3 real-world datasets that we synthetically cast into a total of 66 challenging spectral unmixing benchmarks. We compare our results against a total of 10 baseline methods, including classical methods and a range of learning-based methods. Our results consistently show competitive performance and improved robustness in high noise regimes, when spectra overlap considerably, or when the spectral dimensionality is lowered, making λSplit a new state-of-the-art for spectral unmixing of fluorescent microscopy data. Importantly, λSplit is compatible with spectral data produced by standard confocal microscopes, enabling immediate adoption without specialized hardware modifications.

---

### Swiss-Bench SBP-002: A Frontier Model Comparison on Swiss Legal and Regulatory Tasks

Authors: Fatih Uenal
Published: 2026-03-24
Categories: cs.CL, cs.AI
arXiv: https://arxiv.org/abs/2603.23646v1

While recent work has benchmarked large language models on Swiss legal translation (Niklaus et al., 2025) and academic legal reasoning from university exams (Fan et al., 2025), no existing benchmark evaluates frontier model performance on applied Swiss regulatory compliance tasks. I introduce Swiss-Bench SBP-002, a trilingual benchmark of 395 expert-crafted items spanning three Swiss regulatory domains (FINMA, Legal-CH, EFK), seven task types, and three languages (German, French, Italian), and evaluate ten frontier models from March 2026 using a structured three-dimension scoring framework assessed via a blind three-judge LLM panel (GPT-4o, Claude Sonnet 4, Qwen3-235B) with majority-vote aggregation and weighted kappa = 0.605, with reference answers validated by an independent human legal expert on a 100-item subset (73% rated Correct, 0% Incorrect, perfect Legal Accuracy). Results reveal three descriptive performance clusters: Tier A (35-38% correct), Tier B (26-29%), and Tier C (13-21%). The benchmark proves difficult: even the top-ranked model (Qwen 3.5 Plus) achieves only 38.2% correct, with 47.3% incorrect and 14.4% partially correct. Task type difficulty varies widely: legal translation and case analysis yield 69-72% correct rates, while regulatory Q&A, hallucination detection, and gap analysis remain below 9%. Within this roster (seven open-weight, three closed-source), an open-weight model leads the ranking, and several open-weight models match or outperform their closed-source counterparts. These findings provide an initial empirical reference point for assessing frontier model capability on Swiss regulatory tasks under zero-retrieval conditions.

---

### LLM Inference at the Edge: Mobile, NPU, and GPU Performance Efficiency Trade-offs Under Sustained Load

Authors: Pranay Tummalapalli, Sahil Arayakandy, Ritam Pal, Kautuk Kundan
Published: 2026-03-24
Categories: cs.DC, cs.LG
arXiv: https://arxiv.org/abs/2603.23640v2

Deploying large language models on-device for always-on personal agents demands sustained inference from hardware tightly constrained in power, thermal envelope, and memory. We benchmark Qwen 2.5 1.5B (4-bit quantised) across four platforms: a Raspberry Pi 5 with Hailo-10H NPU, a Samsung Galaxy S24 Ultra, an iPhone 16 Pro, and a laptop NVIDIA RTX 4050 GPU. Using a fixed 258-token prompt over 20 warm-condition iterations per device, we measure throughput, latency, power, and thermal behaviour. For mobile platforms, thermal management supersedes peak compute as the primary constraint: the iPhone 16 Pro loses nearly half its throughput within two iterations, and the S24 Ultra suffers a hard OS-enforced GPU frequency floor that terminates inference entirely. On dedicated hardware, distinct constraints dominate: the RTX 4050 is bounded by its battery power ceiling, while the Hailo-10H is limited by on-module memory bandwidth. The RTX 4050 sustains 131.7 tok/s at 34.1 W; the Hailo-10H sustains 6.9 tok/s at under 2 W with near-zero variance, matching the RTX 4050 in energy proportionality at 19x lower throughput. Results should be interpreted as platform-level deployment characterisations for a single model and prompt type, reflecting hardware and software combined, rather than general claims about hardware capability alone.

---

### Can LLM Agents Be CFOs? Benchmarking Long-Horizon Resource Allocation in an Uncertain Enterprise Environment

Authors: Yi Han, Yan Wang, Lingfei Qian, Haohang Li, Yupeng Cao, Yueru He, Xueqing Peng, Nanhan Shen, Yitao Xu, Yankai Chen, Dongji Feng, Jimin Huang et al.
Published: 2026-03-24
Categories: cs.AI
arXiv: https://arxiv.org/abs/2603.23638v2

Large language model (LLM) agents are increasingly tested on complex tasks, but their ability to allocate scarce resources over long horizons remains unclear. Unlike reactive tasks with immediate feedback, this setting requires agents to make binding commitments under partial observability, delayed consequences, hard resource budgets, and shifting dynamics. We introduce EnterpriseArena, a 132-month CFO simulator that evaluates long-horizon resource allocation under uncertainty in a FinTech lending firm. Agents must manage liquidity, close books, gather costly signals, and request equity or debt financing across changing macroeconomic regimes. The simulator is built from transformed firm-level financial data, anonymized business documents, decade-scale macroeconomic and industry signals, and expert-validated operating rules. Experiments across 23 LLMs and four agent frameworks show that current agents remain far from robust: only 15.4% of trials survive the full horizon, larger models do not reliably outperform smaller ones, and failures cascade across observation, action timing, and capital sizing. These findings establish long-horizon resource allocation under uncertainty as a distinct capability gap for LLM agents.

---

### Stochastic Ray Tracing for the Reconstruction of 3D Gaussian Splatting

Authors: Peiyu Xu, Xin Sun, Krishna Mullia, Raymond Fei, Iliyan Georgiev, Shuang Zhao
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23637v2

Ray-tracing-based 3D Gaussian splatting (3DGS) methods overcome the limitations of rasterization -- rigid pinhole camera assumptions, inaccurate shadows, and lack of native reflection or refraction -- but remain slower due to the cost of sorting all intersecting Gaussians along every ray. Moreover, existing ray-tracing methods still rely on rasterization-style approximations such as shadow mapping for relightable scenes, undermining the generality that ray tracing promises. We present a differentiable, sorting-free stochastic formulation for ray-traced 3DGS -- the first framework that uses stochastic ray tracing to both reconstruct and render standard and relightable 3DGS scenes. At its core is an unbiased Monte Carlo estimator for pixel-color gradients that evaluates only a small sampled subset of Gaussians per ray, bypassing the need for sorting. For standard 3DGS, our method matches the reconstruction quality and speed of rasterization-based 3DGS while substantially outperforming sorting-based ray tracing. For relightable 3DGS, the same stochastic estimator drives per-Gaussian shading with fully ray-traced shadow rays, delivering notably higher reconstruction fidelity than prior work.

---

### Steering Code LLMs with Activation Directions for Language and Library Control

Authors: Md Mahbubur Rahman, Arjun Guha, Harshitha Menon
Published: 2026-03-24
Categories: cs.LG
arXiv: https://arxiv.org/abs/2603.23629v2

Code LLMs often default to particular programming languages and libraries under neutral prompts. We investigate whether these preferences are encoded as approximately linear directions in activation space that can be manipulated at inference time. Using a difference-in-means method, we estimate layer-wise steering vectors for five language/library pairs and add them to model hidden states during generation. Across three open-weight code LLMs, these interventions substantially increase generation toward the target ecosystem under neutral prompts and often remain effective even when prompts explicitly request the opposite choice. Steering strength varies by model and target, with common ecosystems easier to induce than rarer alternatives, and overly strong interventions can reduce output quality. Overall, our results suggest that code-style preferences in LLMs are partly represented by compact, steerable structure in activation space.

---

### Ukrainian Visual Word Sense Disambiguation Benchmark

Authors: Yurii Laba, Yaryna Mohytych, Ivanna Rohulia, Halyna Kyryleyza, Hanna Dydyk-Meush, Oles Dobosevych, Rostyslav Hryniv
Published: 2026-03-24
Categories: cs.CV, cs.AI
arXiv: https://arxiv.org/abs/2603.23627v1

This study presents a benchmark for evaluating the Visual Word Sense Disambiguation (Visual-WSD) task in Ukrainian. The main goal of the Visual-WSD task is to identify, with minimal contextual information, the most appropriate representation of a given ambiguous word from a set of ten images. To construct this benchmark, we followed a methodology similar to that proposed by (CITATION), who previously introduced benchmarks for the Visual-WSD task in English, Italian, and Farsi. This approach allows us to incorporate the Ukrainian benchmark into a broader framework for cross-language model performance comparisons. We collected the benchmark data semi-automatically and refined it with input from domain experts. We then assessed eight multilingual and multimodal large language models using this benchmark. All tested models performed worse than the zero-shot CLIP-based baseline model (CITATION) used by (CITATION) for the English Visual-WSD task. Our analysis revealed a significant performance gap in the Visual-WSD task between Ukrainian and English.

---

### A Theory of LLM Information Susceptibility

Authors: Zhuo-Yang Song, Hua Xing Zhu
Published: 2026-03-24
Categories: cs.LG, cond-mat.stat-mech, cs.AI, cs.CL, nlin.AO
arXiv: https://arxiv.org/abs/2603.23626v1

Large language models (LLMs) are increasingly deployed as optimization modules in agentic systems, yet the fundamental limits of such LLM-mediated improvement remain poorly understood. Here we propose a theory of LLM information susceptibility, centred on the hypothesis that when computational resources are sufficiently large, the intervention of a fixed LLM does not increase the performance susceptibility of a strategy set with respect to budget. We develop a multi-variable utility-function framework that generalizes this hypothesis to architectures with multiple co-varying budget channels, and discuss the conditions under which co-scaling can exceed the susceptibility bound. We validate the theory empirically across structurally diverse domains and model scales spanning an order of magnitude, and show that nested, co-scaling architectures open response channels unavailable to fixed configurations. These results clarify when LLM intervention helps and when it does not, demonstrating that tools from statistical physics can provide predictive constraints for the design of AI systems. If the susceptibility hypothesis holds generally, the theory suggests that nested architectures may be a necessary structural condition for open-ended agentic self-improvement.

---

### Evaluating a Multi-Agent Voice-Enabled Smart Speaker for Care Homes: A Safety-Focused Framework

Authors: Zeinab Dehghani, Rameez Raja Kureshi, Koorosh Aslansefat, Faezeh Alsadat Abedi, Dhavalkumar Thakker, Lisa Greaves, Bhupesh Kumar Mishra, Baseer Ahmad, Tanaya Maslekar
Published: 2026-03-24
Categories: cs.AI, cs.CL
arXiv: https://arxiv.org/abs/2603.23625v1

Artificial intelligence (AI) is increasingly being explored in health and social care to reduce administrative workload and allow staff to spend more time on patient care. This paper evaluates a voice-enabled Care Home Smart Speaker designed to support everyday activities in residential care homes, including spoken access to resident records, reminders, and scheduling tasks. A safety-focused evaluation framework is presented that examines the system end-to-end, combining Whisper-based speech recognition with retrieval-augmented generation (RAG) approaches (hybrid, sparse, and dense). Using supervised care-home trials and controlled testing, we evaluated 330 spoken transcripts across 11 care categories, including 184 reminder-containing interactions. These evaluations focus on (i) correct identification of residents and care categories, (ii) reminder recognition and extraction, and (iii) end-to-end scheduling correctness under uncertainty (including safe deferral/clarification). Given the safety-critical nature of care homes, particular attention is also paid to reliability in noisy environments and across diverse accents, supported by confidence scoring, clarification prompts, and human-in-the-loop oversight. In the best-performing configuration (GPT-5.2), resident ID and care category matching reached 100% (95% CI: 98.86-100), while reminder recognition reached 89.09\% (95% CI: 83.81-92.80) with zero missed reminders (100% recall) but some false positives. End-to-end scheduling via calendar integration achieved 84.65% exact reminder-count agreement (95% CI: 78.00-89.56), indicating remaining edge cases in converting informal spoken instructions into actionable events. The findings suggest that voice-enabled systems, when carefully evaluated and appropriately safeguarded, can support accurate documentation, effective task management, and trustworthy use of AI in care home settings.

---

### Revisiting Real-Time Digging-In Effects: No Evidence from NP/Z Garden-Paths

Authors: Amani Maina-Kilaas, Roger Levy
Published: 2026-03-24
Categories: cs.CL
arXiv: https://arxiv.org/abs/2603.23624v1

Digging-in effects, where disambiguation difficulty increases with longer ambiguous regions, have been cited as evidence for self-organized sentence processing, in which structural commitments strengthen over time. In contrast, surprisal theory predicts no such effect unless lengthening genuinely shifts statistical expectations, and neural language models appear to show the opposite pattern. Whether digging-in is a robust real-time phenomenon in human sentence processing -- or an artifact of wrap-up processes or methodological confounds -- remains unclear. We report two experiments on English NP/Z garden-path sentences using Maze and self-paced reading, comparing human behavior with predictions from an ensemble of large language models. We find no evidence for real-time digging-in effects. Critically, items with sentence-final versus nonfinal disambiguation show qualitatively different patterns: positive digging-in trends appear only sentence-finally, where wrap-up effects confound interpretation. Nonfinal items -- the cleaner test of real-time processing -- show reverse trends consistent with neural model predictions.

---

### M3T: Discrete Multi-Modal Motion Tokens for Sign Language Production

Authors: Alexandre Symeonidis-Herzig, Jianhe Low, Ozge Mercanoglu Sincan, Richard Bowden
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23617v2

Sign language production requires more than hand motion generation. Non-manual features, including mouthings, eyebrow raises, gaze, and head movements, are grammatically obligatory and cannot be recovered from manual articulators alone. Existing 3D production systems face two barriers to integrating them: the body-model fittings they train on retain a facial space too low-dimensional to encode these articulations, and, if richer representations are adopted, standard discrete tokenization suffers from codebook collapse, leaving most of the expression space largely unreachable. We propose SMPL-FX, which couples FLAME's rich expression space with the SMPL-X body parameterisation, and tokenize the resulting representation with modality-specific Finite Scalar Quantization VAEs for body, hands, and face, raising face codebook utilization from 78.8% to 99.0%. M3T is an autoregressive transformer trained on this multi-modal motion vocabulary, with an auxiliary sign-to-text translation objective that encourages semantically grounded embeddings. Across three standard datasets, M3T achieves state-of-the-art sign language production quality, and on NMFs-CSL, where signs are distinguishable only by non-manual features, our model reaches 58.3% accuracy against 49.0% for the strongest comparable baseline, without large-scale sign-language pre-training. Project page: https://cogvis-cvssp.github.io/papers/m3t/

---

### M3T: Discrete Multi-Modal Motion Tokens for Sign Language Production

Authors: Alexandre Symeonidis-Herzig, Jianhe Low, Ozge Mercanoglu Sincan, Richard Bowden
Published: 2026-03-24
Categories: cs.CV
arXiv: https://arxiv.org/abs/2603.23617v1

Sign language production requires more than hand motion generation. Non-manual features, including mouthings, eyebrow raises, gaze, and head movements, are grammatically obligatory and cannot be recovered from manual articulators alone. Existing 3D production systems face two barriers to integrating them: the standard body model provides a facial space too low-dimensional to encode these articulations, and when richer representations are adopted, standard discrete tokenization suffers from codebook collapse, leaving most of the expression space unreachable. We propose SMPL-FX, which couples FLAME's rich expression space with the SMPL-X body, and tokenize the resulting representation with modality-specific Finite Scalar Quantization VAEs for body, hands, and face. M3T is an autoregressive transformer trained on this multi-modal motion vocabulary, with an auxiliary translation objective that encourages semantically grounded embeddings. Across three standard benchmarks (How2Sign, CSL-Daily, Phoenix14T) M3T achieves state-of-the-art sign language production quality, and on NMFs-CSL, where signs are distinguishable only by non-manual features, reaches 58.3% accuracy against 49.0% for the strongest comparable pose baseline.

---

### LLMLOOP: Improving LLM-Generated Code and Tests through Automated Iterative Feedback Loops

Authors: Ravin Ravi, Dylan Bradshaw, Stefano Ruberto, Gunel Jahangirova, Valerio Terragni
Published: 2026-03-24
Categories: cs.SE, cs.AI
arXiv: https://arxiv.org/abs/2603.23613v1

Large Language Models (LLMs) are showing remarkable performance in generating source code, yet the generated code often has issues like compilation errors or incorrect code. Researchers and developers often face wasted effort in implementing checks and refining LLM-generated code, frequently duplicating their efforts. This paper presents LLMLOOP, a framework that automates the refinement of both source code and test cases produced by LLMs. LLMLOOP employs five iterative loops: resolving compilation errors, addressing static analysis issues, fixing test case failures, and improving test quality through mutation analysis. These loops ensure the generation of high-quality test cases that serve as both a validation mechanism and a regression test suite for the generated code. We evaluated LLMLOOP on HUMANEVAL-X, a recent benchmark of programming tasks. Results demonstrate the tool's effectiveness in refining LLM-generated outputs.

---

### LLMORPH: Automated Metamorphic Testing of Large Language Models

Authors: Steven Cho, Stefano Ruberto, Valerio Terragni
Published: 2026-03-24
Categories: cs.SE, cs.AI, cs.CL, cs.LG
arXiv: https://arxiv.org/abs/2603.23611v1

Automated testing is essential for evaluating and improving the reliability of Large Language Models (LLMs), yet the lack of automated oracles for verifying output correctness remains a key challenge. We present LLMORPH, an automated testing tool specifically designed for LLMs performing NLP tasks, which leverages Metamorphic Testing (MT) to uncover faulty behaviors without relying on human-labeled data. MT uses Metamorphic Relations (MRs) to generate follow-up inputs from source test input, enabling detection of inconsistencies in model outputs without the need of expensive labelled data. LLMORPH is aimed at researchers and developers who want to evaluate the robustness of LLM-based NLP systems. In this paper, we detail the design, implementation, and practical usage of LLMORPH, demonstrating how it can be easily extended to any LLM, NLP task, and set of MRs. In our evaluation, we applied 36 MRs across four NLP benchmarks, testing three state-of-the-art LLMs: GPT-4, LLAMA3, and HERMES 2. This produced over 561,000 test executions. Results demonstrate LLMORPH's effectiveness in automatically exposing inconsistencies.

---

### Environment Maps: Structured Environmental Representations for Long-Horizon Agents

Authors: Yenchia Feng, Chirag Sharma, Karime Maamari
Published: 2026-03-24
Categories: cs.AI
arXiv: https://arxiv.org/abs/2603.23610v3

Although large language models (LLMs) have advanced rapidly, robust automation of complex software workflows remains an open problem. In long-horizon settings, agents frequently suffer from cascading errors and environmental stochasticity; a single misstep in a dynamic interface can lead to task failure, resulting in hallucinations or trial-and-error. This paper introduces $\textit{Environment Maps}$: a persistent, agent-agnostic representation that mitigates these failures by consolidating heterogeneous evidence, such as screen recordings and execution traces, into a structured graph. The representation consists of four core components: (1) Contexts (abstracted locations), (2) Actions (parameterized affordances), (3) Workflows (observed trajectories), and (4) Tacit Knowledge (domain definitions and reusable procedures). We evaluate this framework on the WebArena benchmark across five domains. Agents equipped with environment maps achieve a 28.2% success rate, nearly doubling the performance of baselines limited to session-bound context (14.2%) and outperforming agents that have access to the raw trajectory data used to generate the environment maps (23.3%). By providing a structured interface between the model and the environment, Environment Maps establish a persistent foundation for long-horizon planning that is human-interpretable, editable, and incrementally refinable.

---

### Reasoning models do not yet follow their reasoning in autonomous driving: The KITScenes LongTail Dataset

Authors: Royden Wagner, Omer Sahin Tas, Jaime Villa, Felix Hauser, Yinzhe Shen, Marlon Steiner, Dominik Strutz, Carlos Fernandez, Quentin Delfosse, Christoph Weinhuber, Christian Kinzig, Guillermo S. Gutierrez-Cabello et al.
Published: 2026-03-24
Categories: cs.CV, cs.RO
arXiv: https://arxiv.org/abs/2603.23607v3

Handling rare events is the central open challenge in autonomous driving. Reasoning models, which generate explicit chains of reasoning before acting, promise to generalize to such events. Here we show that these models frequently do not follow their own reasoning: the actions they state in their reasoning often diverge from the actions they ultimately execute. We introduce KITScenes LongTail, a curated dataset of rare driving scenarios to quantify this divergence through a measure of semantic reasoning-action coherence. We find that incoherence is widespread across current general-purpose and domain-specific models. Strikingly, when reasoned actions and executed actions disagree, the reasoning is usually right: extracting actions from the reasoning trace and executing them through a kinematic model enforces coherence and typically improves motion planning. These results suggest that the reasoning capabilities of current models exceed what their actions reveal, and that coherently acting on stated reasoning is a prerequisite for trustworthy autonomous driving. Our dataset and evaluation space are available at: https://hf.co/datasets/kit-mrt/kitscenes-longtail

---

### LongTail Driving Scenarios with Reasoning Traces: The KITScenes LongTail Dataset

Authors: Royden Wagner, Omer Sahin Tas, Jaime Villa, Felix Hauser, Yinzhe Shen, Marlon Steiner, Dominik Strutz, Carlos Fernandez, Christian Kinzig, Guillermo S. Guitierrez-Cabello, Hendrik Königshof, Fabian Immel et al.
Published: 2026-03-24
Categories: cs.CV, cs.RO
arXiv: https://arxiv.org/abs/2603.23607v2

In real-world domains such as self-driving, generalization to rare scenarios remains a fundamental challenge. To address this, we introduce a new dataset designed for end-to-end driving that focuses on long-tail driving events. We provide multi-view video data, trajectories, high-level instructions, and detailed reasoning traces, facilitating in-context learning and few-shot generalization. The resulting benchmark for multimodal models, such as VLMs and VLAs, goes beyond safety and comfort metrics by evaluating instruction following and semantic coherence between model outputs. The multilingual reasoning traces in English, Spanish, and Chinese are from domain experts with diverse cultural backgrounds. Thus, our dataset is a unique resource for studying how different forms of reasoning affect driving competence. Our dataset is available at: https://hf.co/datasets/kit-mrt/kitscenes-longtail

---

### LineMVGNN: Anti-Money Laundering with Line-Graph-Assisted Multi-View Graph Neural Networks

Authors: Chung-Hoo Poon, James Kwok, Calvin Chow, Jang-Hyeon Choi
Published: 2026-03-24
Categories: cs.LG, cs.AI, q-fin.CP
arXiv: https://arxiv.org/abs/2603.23584v1

Anti-money laundering (AML) systems are important for protecting the global economy. However, conventional rule-based methods rely on domain knowledge, leading to suboptimal accuracy and a lack of scalability. Graph neural networks (GNNs) for digraphs (directed graphs) can be applied to transaction graphs and capture suspicious transactions or accounts. However, most spectral GNNs do not naturally support multi-dimensional edge features, lack interpretability due to edge modifications, and have limited scalability owing to their spectral nature. Conversely, most spatial methods may not capture the money flow well. Therefore, in this work, we propose LineMVGNN (Line-Graph-Assisted Multi-View Graph Neural Network), a novel spatial method that considers payment and receipt transactions. Specifically, the LineMVGNN model extends a lightweight MVGNN module, which performs two-way message passing between nodes in a transaction graph. Additionally, LineMVGNN incorporates a line graph view of the original transaction graph to enhance the propagation of transaction information. We conduct experiments on two real-world account-based transaction datasets: the Ethereum phishing transaction network dataset and a financial payment transaction dataset from one of our industry partners. The results show that our proposed method outperforms state-of-the-art methods, reflecting the effectiveness of money laundering detection with line-graph-assisted multi-view graph learning. We also discuss scalability, adversarial robustness, and regulatory considerations of our proposed method.

---

### ZeroFold: Protein-RNA Binding Affinity Predictions from Pre-Structural Embeddings

Authors: Josef Hanke, Sebastian Pujalte Ojeda, Shengyu Zhang, Werngard Czechtizky, Leonardo De Maria, Michele Vendruscolo
Published: 2026-03-24
Categories: q-bio.BM, cs.LG
arXiv: https://arxiv.org/abs/2603.23583v1

The accurate prediction of protein-RNA binding affinity remains an unsolved problem in structural biology, limiting opportunities in understanding gene regulation and designing RNA-targeting therapeutics. A central obstacle is the structural flexibility of RNA, as, unlike proteins, RNA molecules exist as dynamic conformational ensembles. Thus, committing to a single predicted structure discards information relevant to binding. Here, we show that this obstacle can be addressed by extracting pre-structural embeddings, which are intermediate representations from a biomolecular foundation model captured before the structure decoding step. Pre-structural embeddings implicitly encode conformational ensemble information without requiring predicted structures. We build ZeroFold, a transformer-based model that combines pre-structural embeddings from Boltz-2 for both protein and RNA molecules through a cross-modal attention mechanism to predict binding affinity directly from sequence. To support training and evaluation, we construct PRADB, a curated dataset of 2,621 unique protein-RNA pairs with experimentally measured affinities drawn from four complementary databases. On a held-out test set constructed with 40% sequence identity thresholds, ZeroFold achieves a Spearman correlation of 0.65, a value approaching the ceiling imposed by experimental measurement noise. Under progressively fairer evaluation conditions that control for training-set overlap, ZeroFold compares favourably with respect to leading structure-based and leading sequence-based predictors, with the performance gap widening as sequence similarity to competitor training data is reduced. These results illustrate how pre-structural embeddings offer a representation strategy for flexible biomolecules, opening a route to affinity prediction for protein-RNA pairs for which no structural data exist.

---

### AI Generalisation Gap In Comorbid Sleep Disorder Staging

Authors: Saswata Bose, Suvadeep Maiti, Shivam Kumar Sharma, Mythirayee S, Tapabrata Chakraborti, Srijitesh Rajendran, Raju S. Bapi
Published: 2026-03-24
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2603.23582v2

Accurate sleep staging is essential for diagnosing OSA and hypopnea in stroke patients. Although PSG is reliable, it is costly, labor-intensive, and manually scored. While deep learning enables automated EEG-based sleep staging in healthy subjects, our analysis shows poor generalization to clinical populations with disrupted sleep. Using Grad-CAM interpretations, we systematically demonstrate this limitation. We introduce iSLEEPS, a newly clinically annotated ischemic stroke dataset (to be publicly released), and evaluate a SE-ResNet plus bidirectional LSTM model for single-channel EEG sleep staging. As expected, cross-domain performance between healthy and diseased subjects is poor. Attention visualizations, supported by clinical expert feedback, show the model focuses on physiologically uninformative EEG regions in patient data. Statistical and computational analyses further confirm significant sleep architecture differences between healthy and ischemic stroke cohorts, highlighting the need for subject-aware or disease-specific models with clinical validation before deployment. A summary of the paper and the code is available at https://himalayansaswatabose.github.io/iSLEEPS_Explainability.github.io/

---

### The Mass Agreement Score: A Point-centric Measure of Cluster Size Consistency

Authors: Randolph Wiredu-Aidoo
Published: 2026-03-24
Categories: stat.ML, cs.LG
arXiv: https://arxiv.org/abs/2603.23581v1

In clustering, strong dominance in the size of a particular cluster is often undesirable, motivating a measure of cluster size uniformity that can be used to filter such partitions. A basic requirement of such a measure is stability: partitions that differ only slightly in their point assignments should receive similar uniformity scores. A difficulty arises because cluster labels are not fixed objects; algorithms may produce different numbers of labels even when the underlying point distribution changes very little. Measures defined directly over labels can therefore become unstable under label-count perturbations. I introduce the Mass Agreement Score (MAS), a point-centric metric bounded in (0, 1) that evaluates the consistency of expected cluster size as measured from the perspective of points in each cluster. Its construction yields fragment robustness by design, assigning similar scores to partitions with similar bulk structure while remaining sensitive to genuine redistribution of cluster mass.

---

### MetaKube: An Experience-Aware LLM Framework for Kubernetes Failure Diagnosis

Authors: Wei Sun, Ting Wang, Xinran Tian, Wanshun Lan, Xuhan Feng, Haoyue Li, Fangxin Wang
Published: 2026-03-24
Categories: cs.LG
arXiv: https://arxiv.org/abs/2603.23580v1

Existing LLM-based Kubernetes diagnostic systems cannot learn from operational experience, operating on static knowledge bases without improving from past resolutions. We present MetaKube, an experience-aware LLM framework through three synergistic innovations: (1) an Episodic Pattern Memory Network (EPMN) that abstracts diagnostic patterns from historical resolutions and provides confidence-calibrated retrieval for both rapid pattern matching and guided causal exploration, (2) a meta-cognitive controller that dynamically routes between intuitive and analytical pathways based on problem familiarity, optimizing the trade-off between speed and depth, and (3) KubeLLM, a locally-deployable 8B model enhanced through domain-specific post-training on our 7,000-sample Kubernetes Fault Resolution Dataset. Evaluation on 1,873 real-world scenarios demonstrates MetaKube transforms Qwen3-8B from 50.9 to 90.5 points, approaching GPT-4.1 performance while ensuring complete data privacy. EPMN contributes 15.3% improvement through experiential learning, with continuous learning experiments showing progressive gains as the system accumulates operational knowledge. The source code and related resources are available at https://github.com/MetaKube-LLM-for-Kubernetes-Diagnosis/MetaKube.

---

### Residual Attention Physics-Informed Neural Networks for Robust Multiphysics Simulation of Steady-State Electrothermal Energy Systems

Authors: Yuqing Zhou, Ze Tao, Fujun Liu
Published: 2026-03-24
Categories: cs.LG, physics.comp-ph
arXiv: https://arxiv.org/abs/2603.23578v1

Efficient thermal management and precise field prediction are critical for the design of advanced energy systems, including electrohydrodynamic transport, microfluidic energy harvesters, and electrically driven thermal regulators. However, the steady-state simulation of these electrothermal coupled multiphysics systems remains challenging for physics-informed neural computation due to strong nonlinear field coupling, temperature-dependent coefficient variability, and complex interface dynamics. This study proposes a Residual Attention Physics-Informed Neural Network (RA-PINN) framework for the unified solution of coupled velocity, pressure, electric-potential, and temperature fields. By integrating a unified five-field operator formulation with residual-connected feature propagation and attention-guided channel modulation, the proposed architecture effectively captures localized coupling structures and steep gradients. We evaluate RA-PINN across four representative energy-relevant benchmarks: constant-coefficient coupling, indirect pressure-gauge constraints, temperature-dependent transport, and oblique-interface consistency. Comparative analysis against Pure-MLP, LSTM-PINN, and pLSTM-PINN demonstrates that RA-PINN achieves superior accuracy, yielding the lowest MSE, RMSE, and relative $L_2$ errors across all scenarios. Notably, RA-PINN maintains high structural fidelity in interface-dominated and variable-coefficient settings where conventional PINN backbones often fail. These results establish RA-PINN as a robust and accurate computational framework for the high-fidelity modeling and optimization of complex electrothermal multiphysics in sustainable energy applications.

---

### The Geometric Price of Discrete Logic: Context-driven Manifold Dynamics of Number Representations

Authors: Long Zhang, Dai-jun Lin, Wei-neng Chen
Published: 2026-03-24
Categories: cs.LG, cs.CL, cs.CY
arXiv: https://arxiv.org/abs/2603.23577v1

Large language models (LLMs) generalize smoothly across continuous semantic spaces, yet strict logical reasoning demands the formation of discrete decision boundaries. Prevailing theories relying on linear isometric projections fail to resolve this fundamental tension. In this work, we argue that task context operates as a non-isometric dynamical operator that enforces a necessary "topological distortion." By applying Gram-Schmidt decomposition to residual-stream activations , we reveal a dual-modulation mechanism driving this process: a class-agnostic topological preservation that anchors global structure to prevent semantic collapse, and a specific algebraic divergence that directionally tears apart cross-class concepts to forge logical boundaries. We validate this geometric evolution across a gradient of tasks, from simple mapping to complex primality testing. Crucially, targeted specific vector ablation establishes a strict causal binding between this topology and model function: algebraically erasing the divergence component collapses parity classification accuracy from 100% to chance levels (38.57%). Furthermore, we uncover a three-phase layer-wise geometric dynamic and demonstrate that under social pressure prompts, models fail to generate sufficient divergence. This results in a "manifold entanglement" that geometrically explains sycophancy and hallucination. Ultimately, our findings revise the linear-isometric presumption, demonstrating that the emergence of discrete logic in LLMs is purchased at an irreducible cost of topological deformation.

---

### Wafer-Level Etch Spatial Profiling for Process Monitoring from Time-Series with Time-LLM

Authors: Hyunwoo Kim, Munyoung Lee, Seung Hyub Jeon, Kyu Sung Lee
Published: 2026-03-24
Categories: stat.AP, cs.AI, cs.LG
arXiv: https://arxiv.org/abs/2603.23576v1

Understanding wafer-level spatial variations from in-situ process signals is essential for advanced plasma etching process monitoring. While most data-driven approaches focus on scalar indicators such as average etch rate, actual process quality is determined by complex two-dimensional spatial distributions across the wafer. This paper presents a spatial regression model that predicts wafer-level etch depth distributions directly from multichannel in-situ process time series. We propose a Time-LLM-based spatial regression model that extends LLM reprogramming from conventional time-series forecasting to wafer-level spatial estimation by redesigning the input embedding and output projection. Using the BOSCH plasma-etching dataset, we demonstrate stable performance under data-limited conditions, supporting the feasibility of LLM-based reprogramming for wafer-level spatial monitoring.

---

### APreQEL: Adaptive Mixed Precision Quantization For Edge LLMs

Authors: Meriem Bouzouad, Yuan-Hao Chang, Jalil Boukhobza
Published: 2026-03-24
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2603.23575v1

Today, large language models have demonstrated their strengths in various tasks ranging from reasoning, code generation, and complex problem solving. However, this advancement comes with a high computational cost and memory requirements, making it challenging to deploy these models on edge devices to ensure real-time responses and data privacy. Quantization is one common approach to reducing memory use, but most methods apply it uniformly across all layers. This does not account for the fact that different layers may respond differently to reduced precision. Importantly, memory consumption and computational throughput are not necessarily aligned, further complicating deployment decisions. This paper proposes an adaptive mixed precision quantization mechanism that balances memory, latency, and accuracy in edge deployment under user-defined priorities. This is achieved by analyzing the layer-wise contribution and by inferring how different quantization types behave across the target hardware platform in order to assign the most suitable quantization type to each layer. This integration ensures that layer importance and the overall performance trade-offs are jointly respected in this design. Our work unlocks new configuration designs that uniform quantization cannot achieve, expanding the solution space to efficiently deploy the LLMs on resource-constrained devices.

---

### PoiCGAN: A Targeted Poisoning Based on Feature-Label Joint Perturbation in Federated Learning

Authors: Tao Liu, Jiguang Lv, Dapeng Man, Weiye Xi, Yaole Li, Feiyu Zhao, Kuiming Wang, Yingchao Bian, Chen Xu, Wu Yang
Published: 2026-03-24
Categories: cs.LG, cs.AI
arXiv: https://arxiv.org/abs/2603.23574v1

Federated Learning (FL), as a popular distributed learning paradigm, has shown outstanding performance in improving computational efficiency and protecting data privacy, and is widely applied in industrial image classification. However, due to its distributed nature, FL is vulnerable to threats from malicious clients, with poisoning attacks being a common threat. A major limitation of existing poisoning attack methods is their difficulty in bypassing model performance tests and defense mechanisms based on model anomaly detection. This often results in the detection and removal of poisoned models, which undermines their practical utility. To ensure both the performance of industrial image classification and attacks, we propose a targeted poisoning attack, PoiCGAN, based on feature-label collaborative perturbation. Our method modifies the inputs of the discriminator and generator in the Conditional Generative Adversarial Network (CGAN) to influence the training process, generating an ideal poison generator. This generator not only produces specific poisoned samples but also automatically performs label flipping. Experiments across various datasets show that our method achieves an attack success rate 83.97% higher than baseline methods, with a less than 8.87% reduction in the main task's accuracy. Moreover, the poisoned samples and malicious models exhibit high stealthiness.

---