3/25/2026
Causal Direct Preference Optimization for Distributionally Robust Generative Recommendation
Causal Direct Preference Optimization method for training LLMs to generate recommendations while mitigating spurious correlations.
3/25/2026
Graphs RAG at Scale: Beyond Retrieval-Augmented Generation With Labeled Property Graphs and Resource Description Framework for Complex and Unknown Search Spaces
Graph RAG framework combining labeled property graphs and RDF for retrieval-augmented generation over structured and semi-structured data.
3/25/2026
T-MAP: Red-Teaming LLM Agents with Trajectory-aware Evolutionary Search
T-MAP uses evolutionary search to red-team LLM agents by exploiting multi-step tool execution vulnerabilities in MCP ecosystems.
3/25/2026
First-Mover Bias in Gradient Boosting Explanations: Mechanism, Detection, and Resolution
Analysis of feature importance bias in gradient boosting models under multicollinearity, affecting SHAP-based explanations.
3/25/2026
WIST: Web-Grounded Iterative Self-Play Tree for Domain-Targeted Reasoning Improvement
WIST framework uses web-grounded iterative self-play with reinforcement learning to improve LLM reasoning in specific domains.
3/25/2026
Early Discoveries of Algorithmist I: Promise of Provable Algorithm Synthesis at Scale
Study on using LLMs for algorithm synthesis with provable guarantees, combining mathematical reasoning with practical performance.
3/25/2026
MCLR: Improving Conditional Modeling in Visual Generative Models via Inter-Class Likelihood-Ratio Maximization and Establishing the Equivalence between Classifier-Free Guidance and Alignment Objectives
Research on improving conditional modeling in diffusion models, establishing equivalence between classifier-free guidance and alignment objectives.
3/25/2026
Reasoner-Executor-Synthesizer: Scalable Agentic Architecture with Static O(1) Context Window
Proposes Reasoner-Executor-Synthesizer architecture for LLM agents that maintains O(1) context window while avoiding hallucination and token cost scaling.
3/25/2026
When Visuals Aren't the Problem: Evaluating Vision-Language Models on Misleading Data Visualizations
Research evaluating Vision-Language Models' ability to detect misleading data visualizations and deceptive captions in charts.
3/25/2026
FAAR: Format-Aware Adaptive Rounding for NVFP4
FAAR quantization method for NVFP4 ultra-low-bit format that adapts rounding to non-uniform numerical grid for efficient LLM edge deployment.
3/25/2026
Rethinking Multimodal Fusion for Time Series: Auxiliary Modalities Need Constrained Fusion
Study on multimodal fusion strategies for time series forecasting showing naive fusion fails and proposing constrained fusion approach.
3/25/2026
Three Creates All: You Only Sample 3 Steps
MTEO method for few-step diffusion sampling by distilling layer-wise, step-wise time embeddings to accelerate inference.
3/25/2026
AI Co-Scientist for Ranking: Discovering Novel Search Ranking Models alongside LLM-based AI Agents with Cloud Computing Access
AI Co-Scientist framework combining LLM agents with cloud computing to automate search ranking research from ideation through GPU training.
3/25/2026
Instruction-Tuned, but Not More Verifiable Instruction-Following: A Cross-Task Diagnosis for LoRA Adapters
Cross-task evaluation study of LoRA adapters showing nominal instruction-tuning labels don't reliably predict realized instruction-following capabilities.
3/25/2026
Symbolic Graph Networks for Robust PDE Discovery from Noisy Sparse Data
Symbolic Graph Network framework for discovering partial differential equations from noisy sparse data without numerical differentiation.
3/25/2026
Learning When to Act: Interval-Aware Reinforcement Learning with Predictive Temporal Structure
Adaptive temporal control system for autonomous agents that learns optimal action intervals using hyperbolic geometry predictive signals.
3/25/2026
CaP-X: A Framework for Benchmarking and Improving Coding Agents for Robot Manipulation
Open-source framework (CaP-X) for benchmarking and improving code-as-policy agents for robot manipulation tasks.
3/25/2026
Sparse but Critical: A Token-Level Analysis of Distributional Shifts in RLVR Fine-Tuning of LLMs
Token-level analysis of distributional shifts in RLVR fine-tuning of LLMs to understand mechanisms underlying reasoning improvements.
3/25/2026
LLM-guided headline rewriting for clickability enhancement without clickbait
LLM-guided headline rewriting system that enhances reader engagement while maintaining editorial integrity and avoiding clickbait.
3/25/2026
Functional Component Ablation Reveals Specialization Patterns in Hybrid Language Model Architectures
Ablation study analyzing specialization patterns in hybrid language models combining attention with state space models on sub-1B parameter models.
3/25/2026
Cognitive Training for Language Models: Towards General Capabilities via Cross-Entropy Games
Framework for building language model general capabilities via automatic curriculum of cross-entropy game tasks for relevant skill discovery.
3/25/2026
Tiny Inference-Time Scaling with Latent Verifiers
Inference-time scaling method using small latent verifiers instead of multimodal LLMs to score and select outputs while reducing computational cost.
3/25/2026
Do Large Language Models Reduce Research Novelty? Evidence from Information Systems Journals
Empirical study measuring semantic novelty of 13,847 IS papers (2020-2025) to assess whether LLM productivity gains translate to genuine intellectual advancement.
3/25/2026
LLMON: An LLM-native Markup Language to Leverage Structure and Semantics at the LLM Interface
LLMON proposes a markup language for LLMs that preserves structure and semantics in prompts, distinguishing between instructions and data in input/output.
3/25/2026
GraphRAG for Engineering Diagrams: ChatP&ID Enables LLM Interaction with P&IDs
ChatP&ID is an agentic RAG framework enabling LLM interaction with engineering P&ID diagrams using knowledge graphs for cost-effective grounded reasoning.
3/25/2026
Ego2Web: A Web Agent Benchmark Grounded in Egocentric Videos
Ego2Web benchmark for multimodal web agents grounded in egocentric video, evaluating agents performing real-world workflows with physical context awareness.
3/25/2026
STRIATUM-CTF: A Protocol-Driven Agentic Framework for General-Purpose CTF Solving
STRIATUM-CTF is an agentic framework using search-based reasoning for automated cybersecurity CTF challenge solving with multi-step stateful reasoning.
3/25/2026
Lie to Me: How Faithful Is Chain-of-Thought Reasoning in Reasoning Models?
Study evaluating faithfulness of chain-of-thought reasoning in LLMs, finding models often produce misleading explanations despite correct outputs.
3/25/2026
flexvec: SQL Vector Retrieval with Programmatic Embedding Modulation
flexvec is a SQL vector retrieval kernel exposing embedding matrices and score arrays for programmatic manipulation by AI agents via Programmatic Embedding Modulation.
3/25/2026
Language Models Can Explain Visual Features via Steering
Method leveraging vision-language models to explain sparse autoencoder features in vision models through causal interventions instead of correlation-based approaches.
3/25/2026
Causal Discovery in Action: Learning Chain-Reaction Mechanisms from Interventions
Theoretical work on causal discovery in chain-reaction systems using interventional data, proving identifiability under cascade-like structural assumptions.
3/25/2026
To Agree or To Be Right? The Grounding-Sycophancy Tradeoff in Medical Vision-Language Models
Evaluation of medical vision-language models revealing a grounding-sycophancy tradeoff, analyzing hallucination and agreement behaviors across six VLMs.
3/25/2026
Toward Faithful Segmentation Attribution via Benchmarking and Dual-Evidence Fusion
Benchmark for evaluating attribution map faithfulness in semantic segmentation models, testing intervention-based faithfulness and perturbation robustness.
3/25/2026
LGSE: Lexically Grounded Subword Embedding Initialization for Low-Resource Language Adaptation
LGSE framework for adapting pretrained language models to low-resource languages using morphologically grounded subword embeddings instead of arbitrary segmentation.
3/25/2026
Learning to Trust: How Humans Mentally Recalibrate AI Confidence Signals
Study examining whether humans can learn to recalibrate AI confidence signals through repeated interaction, testing four calibration conditions with 200 participants.
3/25/2026
AwesomeLit: Towards Hypothesis Generation with Agent-Supported Literature Research
AwesomeLit proposes an agent-supported literature research system for hypothesis generation, designed for inexperienced researchers to identify gaps and propose feasible research directions.
3/25/2026
PopResume: Causal Fairness Evaluation of LLM/VLM Resume Screeners with Population-Representative Dataset
Population-representative resume dataset for causal fairness auditing of LLM/VLM-based screening systems.
3/25/2026
KALAVAI: Predicting When Independent Specialist Fusion Works -- A Quantitative Model for Post-Hoc Cooperative LLM Training
Quantitative model predicting when post-hoc fusion of independent LLM specialists outperforms individual models.
3/25/2026
DALDALL: Data Augmentation for Lexical and Semantic Diverse in Legal Domain by leveraging LLM-Persona
Persona-based data augmentation framework using LLMs for legal domain information retrieval in low-resource settings.
3/25/2026
From Overload to Convergence: Supporting Multi-Issue Human-AI Negotiation with Bayesian Visualization
Bayesian visualization interface supporting multi-issue human-AI negotiation to manage cognitive load.
3/25/2026
From Arithmetic to Logic: The Resilience of Logic and Lookup-Based Neural Networks Under Parameter Bit-Flips
Study of neural network resilience to hardware bit-flip errors, comparing logic-based vs arithmetic architectures.
3/25/2026
KARMA: Knowledge-Action Regularized Multimodal Alignment for Personalized Search at Taobao
LLM fine-tuning framework addressing knowledge-action gap in personalized e-commerce search at Taobao.
3/25/2026
PhotoAgent: A Robotic Photographer with Spatial and Aesthetic Understanding
Embodied AI agent integrating multimodal LLMs with chain-of-thought reasoning for robotic photography tasks.
3/25/2026
Focus, Don't Prune: Identifying Instruction-Relevant Regions for Information-Rich Image Understanding
PinPoint method for identifying instruction-relevant image regions in VLMs to reduce computational overhead.
3/25/2026
When AI Shows Its Work, Is It Actually Working? Step-Level Evaluation Reveals Frontier Language Models Frequently Bypass Their Own Reasoning
Study evaluating whether frontier LLMs genuinely use reasoning steps or generate decorative narratives post-hoc.
3/25/2026
Agent Audit: A Security Analysis System for LLM Agent Applications
Security analysis framework for LLM agent deployments covering model, tool code, credentials, and MCP configurations.
3/25/2026
The Coordinate System Problem in Persistent Structural Memory for Neural Architectures
Dual-View Pheromone Pathway Network (DPPN) architecture for persistent structural memory in neural networks. Identifies coordinate system requirements.
3/25/2026
Agent-Sentry: Bounding LLM Agents via Execution Provenance
Agent-Sentry: Security system for bounding LLM agents via execution provenance tracking. Addresses safety and security concerns in agentic systems.
3/25/2026
Grounding Sim-to-Real Generalization in Dexterous Manipulation: An Empirical Study with Vision-Language-Action Models
Empirical study of sim-to-real transfer for robotic dexterous manipulation using vision-language-action models. Addresses synthetic-to-real gap.
3/25/2026
Confidence Calibration under Ambiguous Ground Truth
Shows that confidence calibration fails when annotator disagreement exists. Proposes calibration against annotator distribution rather than majority labels.