4/24/2026
C-SHAP for time series: An approach to high-level temporal explanations
C-SHAP method for explainability of time series models in high-stakes domains like healthcare and industry.
4/24/2026
Strategic Scaling of Test-Time Compute: A Bandit Learning Approach
Bandit learning approach for adaptive test-time compute allocation to LLM queries based on estimated difficulty rather than uniform scaling.
4/24/2026
KompeteAI: Accelerated Autonomous Multi-Agent System for End-to-End Pipeline Generation for Machine Learning Problems
KompeteAI autonomous multi-agent system for ML pipeline generation addressing exploration and execution bottlenecks in LLM-based AutoML.
4/24/2026
PosterForest: Hierarchical Multi-Agent Collaboration for Scientific Poster Generation
PosterForest multi-agent system for hierarchical scientific poster generation using tree-structured planning for content and layout organization.
4/24/2026
Learning Reasoning Reward Models from Expert Demonstration via Inverse Reinforcement Learning
AIRL framework learning reasoning reward models from expert demonstrations for LLM improvement, bridging supervised fine-tuning and outcome-based RL.
4/24/2026
Speculative Actions: A Lossless Framework for Faster Agentic Systems
Speculative Actions framework for accelerating AI agents through parallel action generation, reducing sequential API latency in interactive environments.
4/24/2026
OpenEstimate: Evaluating LLMs on Reasoning Under Uncertainty with Real-World Data
OpenEstimate benchmark evaluating LLMs on reasoning under uncertainty using real-world data from healthcare, finance, and knowledge work domains.
4/24/2026
ReProbe: Efficient Test-Time Scaling of Multi-Step Reasoning by Probing Internal States of Large Language Models
ReProbe method for efficient test-time scaling of LLM reasoning by probing internal states instead of expensive process reward models.
4/24/2026
The Specification Trap: Why Static Value Alignment Alone Is Insufficient for Robust Alignment
Philosophical analysis arguing static value alignment approaches are insufficient for AI systems under capability scaling and distributional shift.
4/24/2026
AgencyBench: Benchmarking the Frontiers of Autonomous Agents in 1M-Token Real-World Contexts
arXiv benchmark (AgencyBench) evaluating LLM agents on long-horizon real-world tasks with 1M-token context and automated evaluation.
4/24/2026
AgentDoG: A Diagnostic Guardrail Framework for AI Agent Safety and Security
arXiv framework (AgentDoG) for diagnosing and mitigating safety/security risks in autonomous AI agents with unified taxonomy.
4/24/2026
Grounding Machine Creativity in Game Design Knowledge Representations: Empirical Probing of LLM-Based Executable Synthesis of Goal Playable Patterns under Structural Constraints
arXiv paper on LLM-based synthesis of game design patterns and executable gameplay code using design knowledge representations.
4/24/2026
Empirical Comparison of Agent Communication Protocols for Task Orchestration
arXiv benchmark comparing LLM agent communication protocols for task orchestration, tool integration, and multi-agent delegation.
4/24/2026
On the Relationship between Bayesian Networks and Probabilistic Structural Causal Models
arXiv paper analyzing mathematical relationships between Bayesian networks and structural causal models with uncertainty.
4/24/2026
Wiring the 'Why': A Unified Taxonomy and Survey of Abductive Reasoning in LLMs
arXiv survey of abductive reasoning in LLMs, covering inference of plausible explanations across diverse task settings.
4/24/2026
DRBENCHER: Can Your Agent Identify the Entity, Retrieve Its Properties and Do the Math?
arXiv benchmark (DRBENCHER) for evaluating deep research agents combining web browsing with multi-step computation tasks.
4/24/2026
QuarkMedSearch: A Long-Horizon Deep Search Agent for Exploring Medical Intelligence
arXiv paper on QuarkMedSearch, an agentic AI system for medical domain search with multi-hop data construction and evaluation.
4/24/2026
AISafetyBenchExplorer: A Metric-Aware Catalogue of AI Safety Benchmarks Reveals Fragmented Measurement and Weak Benchmark Governance
arXiv catalog of 195 AI safety benchmarks with structured metadata revealing fragmentation in LLM safety measurement practices.
4/24/2026
HWE-Bench: Benchmarking LLM Agents on Real-World Hardware Bug Repair Tasks
arXiv benchmark (HWE-Bench) for evaluating LLM agents on real-world hardware bug repair at repository scale with 417 task instances.
4/24/2026
ReactBench: A Benchmark for Topological Reasoning in MLLMs on Chemical Reaction Diagrams
arXiv benchmark (ReactBench) evaluating multimodal LLMs on topological reasoning over complex chemical reaction diagrams.
4/24/2026
Beyond One Output: Visualizing and Comparing Distributions of Language Model Generations
arXiv paper proposing visualization methods for examining distributions of LLM outputs instead of single samples.
4/24/2026
FSFM: A Biologically-Inspired Framework for Selective Forgetting of Agent Memory
FSFM proposes biologically-inspired selective forgetting mechanism for LLM agents based on hippocampal consolidation and forgetting curves for memory management.
4/24/2026
Reinforcement Learning with Foundation Priors: Let the Embodied Agent Efficiently Learn on Its Own
Framework leveraging foundation model priors to enable efficient reinforcement learning for robotic manipulation tasks with minimal environment interaction.
4/24/2026
Preserving Knowledge in Large Language Model with Model-Agnostic Self-Decompression
Model-Agnostic Self-Decompression prevents catastrophic forgetting in LLMs during fine-tuning and addresses performance degradation in multimodal models.
4/24/2026
Federated Co-tuning Framework for Large and Small Language Models
FedCoLLM proposes federated parameter-efficient framework for co-tuning large and small language models with mutual enhancement.
4/24/2026
Analytical FFN-to-MoE Restructuring via Activation Pattern Analysis
Analytical FFN-to-MoE Restructuring converts dense LLM feed-forward networks to Mixture-of-Experts without extensive retraining, reducing inference costs.
4/24/2026
SafeMERGE: Preserving Safety Alignment in Fine-Tuned Large Language Models via Selective Layer-Wise Model Merging
SafeMERGE preserves safety alignment during LLM fine-tuning via selective layer-wise model merging, maintaining task utility without custom algorithms.
4/24/2026
Logic Jailbreak: Efficiently Unlocking LLM Safety Restrictions Through Formal Logical Expression
LogiBreak presents black-box jailbreak method using logical expression translation to bypass LLM safety mechanisms, analyzing distributional discrepancies in alignment.
4/24/2026
Secure LLM Fine-Tuning via Safety-Aware Probing
Secure LLM Fine-Tuning via Safety-Aware Probing addresses how fine-tuning compromises LLM safety alignment and proposes detection methods.
4/24/2026
Counterfactual Segmentation Reasoning: Diagnosing and Mitigating Pixel-Grounding Hallucination
Counterfactual Segmentation Reasoning method diagnoses and mitigates pixel-grounding hallucinations in Vision-Language Models for segmentation tasks.
4/24/2026
mGRADE: Minimal Recurrent Gating Meets Delay Convolutions for Lightweight Sequence Modeling
mGRADE proposes lightweight sequence modeling architecture balancing local and global context under memory constraints for edge devices.
4/24/2026
Algebraic Language Models for Inverse Design of Metamaterials via Diffusion Transformers
DiffuMeta combines diffusion transformers with algebraic language models for inverse design of 3D metamaterials, addressing computational complexity in material discovery.
4/24/2026
A Comprehensive Guide to Differential Privacy: From Theory to User Expectations
Comprehensive survey of differential privacy theory, applications, and gap between formal guarantees and user expectations in ML and data security contexts.
4/24/2026
HyperAdapt: Simple High-Rank Adaptation
HyperAdapt introduces a parameter-efficient fine-tuning method that reduces trainable parameters while adapting foundation models to specialized tasks with lower memory and compute requirements.
4/24/2026
InfiniPipe: Elastic Pipeline Parallelism for Efficient Variable-Length Long-Context LLM Training
InfiniPipe proposes elastic pipeline parallelism to reduce communication overhead in long-context LLM training, addressing memory and efficiency tradeoffs between sequence and token-level parallelism.
4/24/2026
Context Is What You Need: The Maximum Effective Context Window for Real World Limits of LLMs
Defines effective context window for LLMs and creates standardized testing method across models and problem types.
4/24/2026
Geo-R1: Improving Few-Shot Geospatial Referring Expression Understanding with Reinforcement Fine-Tuning
Reinforcement fine-tuning for geospatial referring expression understanding with limited labeled data.
4/24/2026
Compose and Fuse: Revisiting the Foundational Bottlenecks in Multimodal Reasoning
Studies cross-modal reasoning in MLLMs and when modality interactions help or harm performance.
4/24/2026
ChessArena: A Chess Testbed for Evaluating Strategic Reasoning Capabilities of Large Language Models
Chess-based testbed for evaluating strategic reasoning and rule adherence in LLMs.
4/24/2026
Towards Multimodal Active Learning: Efficient Learning with Limited Paired Data
Framework for active learning with unaligned multimodal data to reduce annotation costs.
4/24/2026
Federated Learning for Surgical Vision in Appendicitis Classification: Results of the FedSurg EndoVis 2024 Challenge
First federated learning benchmark for surgical video analysis without sharing patient data.
4/24/2026
Accelerating Vision Transformers with Adaptive Patch Sizes
Adaptive patch sizing for Vision Transformers reduces tokens via content-aware allocation.
4/24/2026
Mitigating Lost in Multi-turn Conversation via Curriculum RL with Verifiable Accuracy and Abstention Rewards
Curriculum RL framework addressing performance degradation in multi-turn conversations with verifiable rewards.
4/24/2026
Fairness Evaluation and Inference Level Mitigation in LLMs
Inference-level mitigation techniques for fairness issues in LLMs without expensive retraining.
4/24/2026
RELOOP: Recursive Retrieval with Multi-Hop Reasoner and Planners for Heterogeneous QA
RAG framework using hierarchical sequences for multi-hop reasoning over heterogeneous sources (documents, tables, KGs).
4/24/2026
When to Trust the Answer: Question-Aligned Semantic Nearest Neighbor Entropy for Safer Surgical VQA
Uncertainty estimation method for surgical VQA systems accounting for question context in safety-critical applications.
4/24/2026
Retrofit: Continual Learning with Controlled Forgetting for Binary Security Detection and Analysis
Continual learning framework for binary malware detection with controlled forgetting to adapt as threats evolve.
4/24/2026
VVS: Accelerating Speculative Decoding for Visual Autoregressive Generation via Partial Verification Skipping
Accelerates visual autoregressive generation via partial verification skipping in speculative decoding.
4/24/2026
Why Do Language Model Agents Whistleblow?
Studies why LLM agents disclose information to external parties contrary to user instructions, examining alignment behavior.
4/24/2026
Musical Score Understanding Benchmark: Evaluating Large Language Models' Comprehension of Complete Musical Scores
Benchmark for evaluating LLM and vision-language model comprehension of musical scores in ABC and PDF formats.