6/1/2026
PMF-CL: Pareto-Minimal-Forgetting Continual Learner for Conflicting Tasks
PMF-CL algorithm for continual learning using Pareto-minimal-forgetting approach to minimize catastrophic forgetting on conflicting tasks.
6/1/2026
Efficient Learning of Deep State Space Models via Importance Smoothing
Importance smoothing technique for training deep state space models at scale, combining variational and backpropagation approaches.
6/1/2026
PRISM: Preference-Aware Influence Function Based Data Selection Method for Efficient Fine-Tuning
PRISM data selection method using preference-aware influence functions to select high-value examples for efficient LLM fine-tuning.
6/1/2026
Agent JIT Compilation for Latency-Optimizing Web Agent Planning and Scheduling
Agent JIT compilation method for web agents that reduces latency by batching multiple tool calls before LLM evaluation.
6/1/2026
The Distillation Game: Adaptive Attacks & Efficient Defenses
Game-theoretic framework analyzing trade-off between model utility and distillation attacks with adaptive student and teacher defenses.
6/1/2026
Learned Relay Representations for Forward-Thinking Discrete Diffusion Models
Learned Relay Representations method for masked diffusion models to preserve internal computation between denoising iterations.
6/1/2026
GEM: Geometric Entropy Mixing for Optimal LLM Data Curation
GEM framework for LLM pre-training data curation using geometric entropy mixing on hypersphere to optimize data composition.
6/1/2026
Conveyance: A Versatile Framework for Learning in Structured Class Spaces
Framework for structured classification using loss functions that exploit class relationships. Addresses structure-agnostic loss limitations.
6/1/2026
Inferring the Size of Large Language Models From Popular Text Memorization
Black-box method to infer lower bounds on LLM parameter counts from text memorization patterns without model access. Novel inference technique.
6/1/2026
Deep Adaptive Dimension Reduction for Bayesian Inference in Inverse Problems
Dimension reduction framework for Bayesian inference in high-dimensional PDE inverse problems using variational flows. Specialized mathematical methods.
6/1/2026
ProofWala: A Framework for Multilingual Proof Data Synthesis and Theorem-Proving
Framework for multilingual proof data synthesis and theorem proving. Infrastructure for interfacing theorem provers at repository scale with parallel execution.
6/1/2026
Chain-of-Thought Reasoning In The Wild Is Not Always Faithful
Research showing chain-of-thought reasoning in LLMs exhibits unfaithfulness on natural prompts without explicit biases, revealing reasoning discrepancies.
6/1/2026
Unraveling LoRA Interference: Orthogonal Subspaces for Robust Model Merging
Research on LoRA interference in model merging using orthogonal subspaces. Addresses performance degradation when combining multiple fine-tuned LMs.
6/1/2026
Esoteric Language Models: A Family of Any-Order Diffusion LLMs
Eso-LMs: Family of diffusion language models interpolating autoregressive and masked diffusion paradigms with parallel generation and KV caching.
6/1/2026
Position: Quantum Kernel Machines Should Move Beyond Scalar-Valued Kernels to Realize Their Potential
Position paper arguing quantum kernel machines should explore matrix-valued kernels beyond scalar kernels for advantages.
6/1/2026
DISCO: Mitigating Bias in Deep Learning with Conditional Distance Correlation
DISCO method uses conditional distance correlation to mitigate dataset bias in deep learning through causal framework.
6/1/2026
Conformal C2ST: Turning weak classifiers into strong two-sample tests
Conformal C2ST method improves two-sample testing by transforming weak classifiers into strong statistical tests.
6/1/2026
Benchmarking Uncertainty and its Disentanglement in multi-label Chest X-Ray Classification
Benchmarking uncertainty quantification methods in chest X-ray classification for trustworthy medical AI deployment.
6/1/2026
Stochastic Gradients under Nuisances
Theoretical analysis of stochastic gradient optimization with unknown nuisance parameters, establishing convergence guarantees.
6/1/2026
Neuro-Symbolic Predictive Process Monitoring
Neuro-symbolic approach integrating deep learning with temporal logic for business process monitoring and suffix prediction.
6/1/2026
MatchFixAgent: Language-Agnostic Autonomous Repository-Level Code Translation Validation and Repair
MatchFixAgent: Language-agnostic AI agent for autonomous code translation validation and repair across programming languages.
6/1/2026
Unmute the Patch Tokens: Rethinking Probing in Multi-Label Audio Classification
Research on audio classification showing global pooling creates bottleneck in frozen model probing; proposes using patch tokens instead.
6/1/2026
ReTabAD: A Benchmark for Restoring Semantic Context in Tabular Anomaly Detection
arXiv benchmark for tabular anomaly detection with semantic context and domain knowledge metadata.
6/1/2026
Adversarial Robustness in One-Stage Learning-to-Defer
arXiv research on adversarial robustness in one-stage learning-to-defer systems for hybrid predictor-expert decision-making.
6/1/2026
CaptionFormer: Unified Segmentation, Tracking, and Captioning for Spatio-Temporal Objects
arXiv paper on unified spatio-temporal video understanding combining object detection, tracking, and natural language captioning.
6/1/2026
Non-Parametric Probabilistic Robustness: A Conservative Risk Estimator under Unknown Perturbation Distributions
arXiv paper on probabilistic robustness evaluation for deep learning models under unknown perturbation distributions.
6/1/2026
Outcome-Aware Spectral Feature Learning for Instrumental Variable Regression
arXiv research on instrumental variable regression using outcome-aware spectral features for causal effect estimation.
6/1/2026
Conditional Coverage Diagnostics for Conformal Prediction
arXiv research on evaluating conditional coverage in conformal prediction methods for reliable predictive systems.
6/1/2026
Agentic Physical AI toward a Domain-Specific Foundation Model for Energy Systems: A Case Study on Nuclear Reactor Control
arXiv paper on domain-specific foundation models for agentic physical AI, case study on nuclear reactor control with safety constraints.
6/1/2026
Regret-Based Federated Causal Discovery with Unknown Interventions
arXiv research on federated causal discovery handling heterogeneous client models with unknown interventions.
6/1/2026
Rethinking Multimodal Few-Shot 3D Point Cloud Segmentation: From Fused Refinement to Decoupled Arbitration
arXiv paper on few-shot 3D point cloud segmentation using decoupled expert approach for multimodal data.
6/1/2026
Who Endorsed It? Measuring Authority Bias Across Expertise Levels in Language Models
arXiv study measuring authority bias in language models across mathematical, legal, and medical reasoning tasks with endorsement sources.
6/1/2026
Performance and Complexity Trade-off Optimization of Speech Models During Training
arXiv research on optimizing neural network layer sizes for speech models, addressing performance-complexity trade-offs during training.
6/1/2026
SERA: Soft-Verified Efficient Repository Agents
SERA framework for efficiently training open-weight repository-specific coding agents with soft verification without expensive full training.
6/1/2026
Token Sparse Attention: Efficient Long-Context Inference with Interleaved Token Selection
Token Sparse Attention method for efficient long-context LLM inference using dynamic interleaved token selection per layer/head.
6/1/2026
BAT: Better Audio Transformer Guided by Convex Gated Probing
BAT uses convex gated probing mechanism to evaluate audio SSL embeddings better than finetuning on AudioSet.
6/1/2026
Quantifying Error Propagation and Model Collapse in Diffusion Models
Theoretical analysis of error propagation and model collapse when training diffusion models recursively on synthetic data.
6/1/2026
IAPO: Information-Aware Policy Optimization for Token-Efficient Reasoning
IAPO framework uses information-theoretic token-wise rewards for efficient reasoning in LLMs, reducing inference token costs.
6/1/2026
KernelCraft: Benchmarking for Agentic Close-to-Metal Kernel Generation on Emerging Hardware
KernelCraft benchmarks agentic LLM systems for generating low-level kernels on emerging AI accelerators with novel ISAs.
6/1/2026
GradMem: Learning to Write Context into Memory with Test-Time Gradient Descent
GradMem method using test-time gradient descent for compressive memory conditioning in LLMs, reducing KV-cache memory overhead.
6/1/2026
Learning-to-Defer with Expert-Conditional Advice
Learning-to-defer framework allowing dynamic selection of expert-specific information like retrieved documents and tool outputs.
6/1/2026
IsoCLIP: Decomposing CLIP Projectors for Efficient Intra-modal Alignment
Method decomposing CLIP projectors to improve intra-modal alignment for image-to-image and text-to-text retrieval tasks.
6/1/2026
SimulCost: A Cost-Aware Benchmark and Toolkit for Automating Physics Simulations with LLMs
Benchmark and toolkit for evaluating LLM agents on physics simulations with cost-aware metrics accounting for tool-use expenses.
6/1/2026
Beyond Augmented-Action Surrogates for Multi-Expert Learning-to-Defer
Learning-to-defer system framework for routing inputs between classifier and multiple experts with theoretical improvements.
6/1/2026
Learning to Reason with Insight for Informal Theorem Proving
DeepInsight method for informal theorem proving with LLMs by identifying core solution techniques through insight recognition.
6/1/2026
Aligning Dense Retrievers with LLM Utility via Distillation
Utility-Aligned Embeddings framework combining dense retrieval with LLM utility for improved RAG performance and computational efficiency.
6/1/2026
Inference of Online Newton Methods with Nesterov's Accelerated Sketching
Research on uncertainty quantification for online optimization methods using sketching to reduce computational complexity.
6/1/2026
Self-Captioning Multimodal Interaction Tuning: Amplifying Exploitable Redundancies for Robust Vision Language Models
Method for improving vision-language model robustness by exploiting multimodal redundancies to reduce hallucination on corrupted inputs.
6/1/2026
Online Learning-to-Defer with Varying Experts
Online learning-to-defer algorithm for routing queries between models and varying expert pools with bandit feedback.
6/1/2026
ASH: Agents that Self-Hone via Embodied Learning
ASH: agentic system for embodied policy learning from unlabeled internet video using self-improvement loop with inverse dynamics models.