4/3/2026
Universal Hypernetworks for Arbitrary Models
ArXiv paper introducing Universal Hypernetworks that generate weights for arbitrary model architectures using descriptors.
4/3/2026
Smoothing the Landscape: Causal Structure Learning via Diffusion Denoising Objectives
ArXiv paper applying diffusion denoising objectives to causal structure learning from observational data.
4/3/2026
Model-Based Reinforcement Learning for Control under Time-Varying Dynamics
ArXiv paper on model-based reinforcement learning for control systems with time-varying dynamics.
4/3/2026
SKILL0: In-Context Agentic Reinforcement Learning for Skill Internalization
ArXiv paper on in-context agentic reinforcement learning enabling LLM agents to internalize skills at inference time.
4/3/2026
Crystalite: A Lightweight Transformer for Efficient Crystal Modeling
ArXiv paper on lightweight diffusion transformer for crystal structure generation using subatomic tokenization.
4/3/2026
Unifying Group-Relative and Self-Distillation Policy Optimization via Sample Routing
ArXiv paper unifying group-relative and self-distillation policy optimization for LLM post-training with improved credit assignment.
4/3/2026
Taming the Exponential: A Fast Softmax Surrogate for Integer-Native Edge Inference
ArXiv paper proposing Head-Calibrated Clipped-Linear Softmax as efficient surrogate for attention softmax in edge inference.
4/3/2026
go-$m$HC: Direct Parameterization of Manifold-Constrained Hyper-Connections via Generalized Orthostochastic Matrices
ArXiv paper on exact parameterization of doubly stochastic matrices for learned mixing in neural networks.
4/3/2026
Batched Contextual Reinforcement: A Task-Scaling Law for Efficient Reasoning
Single-stage training paradigm for efficient LLM reasoning that reduces token consumption in chain-of-thought without degrading quality.
4/3/2026
A Learning-Based Cooperative Coevolution Framework for Heterogeneous Large-Scale Global Optimization
Learning-based cooperative coevolution framework addressing heterogeneous large-scale global optimization via adaptive low-dimensional optimizers.
4/3/2026
Bias Inheritance in Neural-Symbolic Discovery of Constitutive Closures Under Function-Class Mismatch
Neural-symbolic framework for discovering constitutive closures in nonlinear PDEs from spatiotemporal data while avoiding spurious physical recovery.
4/3/2026
Regularizing Attention Scores with Bootstrapping
Research on regularizing attention scores in vision transformers using bootstrapping to improve interpretability and reduce noisy attention maps.
4/3/2026
Safety, Security, and Cognitive Risks in World Models
Analysis of safety, security, and cognitive risks in world models used for autonomous decision-making in robotics, autonomous vehicles, and agentic AI systems.
4/3/2026
When AI Gets it Wong: Reliability and Risk in AI-Assisted Medication Decision Systems
Study of reliability gaps in AI-assisted medication systems, highlighting risks in healthcare decision support.
4/3/2026
Infeasibility Aware Large Language Models for Combinatorial Optimization
Framework combining LLMs with infeasibility detection for NP-hard combinatorial optimization problems.
4/3/2026
Efficient Equivariant Transformer for Self-Driving Agent Modeling
Equivariant transformer architecture for modeling agent behaviors in autonomous driving with SE(2) symmetry.
4/3/2026
The Newton-Muon Optimizer
New optimizer deriving design principles from Muon, improving LLM training efficiency through surrogate model analysis.
4/3/2026
Prime Once, then Reprogram Locally: An Efficient Alternative to Black-Box Service Model Adaptation
Method for efficiently adapting closed-box LLM APIs to target tasks by priming followed by local optimization.
4/3/2026
ProdCodeBench: A Production-Derived Benchmark for Evaluating AI Coding Agents
Benchmark dataset for evaluating AI coding agents based on production workloads, addressing language distribution and codebase structure gaps.
4/3/2026
Does Your Optimizer Care How You Normalize? Normalization-Optimizer Coupling in LLM Training
Study of interactions between normalization methods and optimizers in LLM training at 1B parameters.
4/3/2026
LiteInception: A Lightweight and Interpretable Deep Learning Framework for General Aviation Fault Diagnosis
LiteInception: lightweight interpretable deep learning framework for fault diagnosis on edge devices.
4/3/2026
LiveMathematicianBench: A Live Benchmark for Mathematician-Level Reasoning with Proof Sketches
LiveMathematicianBench: benchmark for evaluating LLM mathematical reasoning capabilities with proof sketches.
4/3/2026
Language-Pretraining-Induced Bias: A Strong Foundation for General Vision Tasks
Study on language pre-training bias improving performance on general vision tasks through cross-modality transfer.
4/3/2026
Investigating Permutation-Invariant Discrete Representation Learning for Spatially Aligned Images
Analysis of permutation-invariant discrete representation learning for spatially aligned images using vector quantization.
4/3/2026
Woosh: A Sound Effects Foundation Model
Woosh: open-source sound effects foundation model from Sony AI with architecture, training details, and benchmarks.
4/3/2026
Homogenized Transformers
Theoretical analysis of multi-head self-attention transformers using particle systems and homogenization limits.
4/3/2026
Curia-2: Scaling Self-Supervised Learning for Radiology Foundation Models
Curia-2 foundation model using self-supervised learning for medical imaging analysis on CT and MRI data.
4/3/2026
Systematic Analyses of Reinforcement Learning Controllers in Signalized Urban Corridors
Comparison of centralized and decentralized RL controllers for traffic signal control in urban corridor networks.
4/3/2026
Mining Instance-Centric Vision-Language Contexts for Human-Object Interaction Detection
Mining instance-centric vision-language contexts for human-object interaction detection, leveraging VLMs to improve semantic understanding and contextual reasoning.
4/3/2026
LatentUM: Unleashing the Potential of Interleaved Cross-Modal Reasoning via a Latent-Space Unified Model
LatentUM unified model for interleaved cross-modal reasoning combining visual understanding, generation, and world dynamics in latent space.
4/3/2026
Intelligent Cloud Orchestration: A Hybrid Predictive and Heuristic Framework for Cost Optimization
Hybrid framework combining LSTM workload prediction with game-theoretic heuristics for cloud cost optimization during dynamic workload changes.
4/3/2026
AstroConcepts: A Large-Scale Multi-Label Classification Corpus for Astrophysics
AstroConcepts corpus of 21,702 astrophysics abstracts for multi-label classification research addressing extreme class imbalance with specialized terminology.
4/3/2026
Do Lexical and Contextual Coreference Resolution Systems Degrade Differently under Mention Noise? An Empirical Study on Scientific Software Mentions
Shared task participation comparing lexical and contextual approaches for cross-document software mention coreference resolution under mention noise.
4/3/2026
The Expert Strikes Back: Interpreting Mixture-of-Experts Language Models at Expert Level
Analysis of Mixture-of-Experts LLM interpretability at expert level, comparing sparsity properties to dense feed-forward networks using probing methods.
4/3/2026
Computing the Exact Pareto Front in Average-Cost Multi-Objective Markov Decision Processes
Characterization of exact Pareto fronts in average-cost multi-objective MDPs, extending prior work on discounted settings.
4/3/2026
From High-Dimensional Spaces to Verifiable ODD Coverage for Safety-Critical AI-based Systems
Method for verifying operational design domain coverage for safety-critical AI systems in aviation, addressing EASA certification requirements.
4/3/2026
When to ASK: Uncertainty-Gated Language Assistance for Reinforcement Learning
ASK framework combines smaller language models with RL policies to enhance out-of-distribution generalization by gating LM assistance based on agent uncertainty.
4/3/2026
(PAC-)Learning state machines from data streams: A generic strategy and an improved heuristic (Extended version)
Extended publication on learning state machines from data streams with PAC-bounds analysis and heuristic improvements.
4/3/2026
BVFLMSP : Bayesian Vertical Federated Learning for Multimodal Survival with Privacy
Bayesian vertical federated learning approach for multimodal survival prediction with privacy preservation and uncertainty quantification.
4/3/2026
De Jure: Iterative LLM Self-Refinement for Structured Extraction of Regulatory Rules
De Jure is an automated pipeline for extracting structured regulatory rules from legal documents using LLM self-refinement, requiring no human annotation or domain-specific data.
4/3/2026
Grounded Token Initialization for New Vocabulary in LMs for Generative Recommendation
Analysis of token initialization strategies for new vocabulary in language models used for generative recommendation systems.
4/3/2026
Transformers Can Solve Non-Linear and Non-Markovian Filtering Problems in Continuous Time For Conditionally Gaussian Signals
Theoretical analysis showing transformers can solve non-linear non-Markovian stochastic filtering for conditionally Gaussian signals.
4/3/2026
Moonwalk: Inverse-Forward Differentiation
Moonwalk: inverse-forward differentiation technique addressing backpropagation's memory requirements for training deeper networks.
4/3/2026
A Comprehensive Graph Pooling Benchmark: Effectiveness, Robustness and Generalizability
Comprehensive benchmark of 17 graph pooling methods across 28 datasets evaluating effectiveness, robustness and generalization.
4/3/2026
In-context Learning in Presence of Spurious Correlations
Study of in-context learning in LLMs with spurious correlations, examining transformer robustness to spurious features in classification.
4/3/2026
Automatic selection of the best neural architecture for time series forecasting
Automated neural architecture selection for time series forecasting comparing LSTMs, GRUs, Transformers, and State-Space Models.
4/3/2026
Fragility-aware Classification for Understanding Risk and Improving Generalization
ML research on classification metrics accounting for confidence in incorrect predictions for safety-critical applications.
4/3/2026
A Simultaneous Approach for Training Neural Differential-Algebraic Systems of Equations
Scientific ML research on training neural differential-algebraic equation systems extending neural ODE methods.
4/3/2026
GradPower: Powering Gradients for Faster Language Model Pre-Training
GradPower: lightweight gradient transformation technique for accelerating LLM pre-training via sign-power transformation, requiring minimal code changes.
4/3/2026
Monotone and Conservative Policy Iteration Beyond the Tabular Case
ML research introducing Reliable Policy Iteration variants that maintain theoretical guarantees under function approximation in reinforcement learning.