3/20/2026
Soft-Di[M]O: Improving One-Step Discrete Image Generation with Soft Embeddings
One-step image generation improvement using soft embeddings in distilled masked diffusion models, enabling gradient flow for post-distillation refinements.
3/20/2026
Blind to Position, Biased in Language: Probing Mid-Layer Representational Bias in Vision-Language Encoders for Zero-Shot Language-Grounded Spatial Understanding
Analysis of positional and language bias in mid-layer representations of vision-language encoders for zero-shot language-grounded spatial understanding.
3/20/2026
Redundancy-as-Masking: Formalizing the Artificial Age Score (AAS) to Model Memory Aging in Generative AI
Artificial Age Score framework formalizing memory aging in LLMs, modeling how semantic and episodic information degrades across conversational sessions.
3/20/2026
VeriEquivBench: An Equivalence Score for Ground-Truth-Free Evaluation of Formally Verifiable Code
Benchmark using equivalence scoring for ground-truth-free evaluation of formally verifiable code generated by LLMs in languages like Dafny.
3/20/2026
AdaSwitch: Balancing Exploration and Guidance in Knowledge Distillation via Adaptive Switching
Knowledge distillation method for small language models that balances exploration and guidance through adaptive switching to address exposure bias.
3/20/2026
Evaluating Hallucinations in Audio-Visual Multimodal LLMs with Spoken Queries under Diverse Acoustic Conditions
Benchmark evaluating hallucinations in audio-visual multimodal LLMs with spoken queries under diverse acoustic conditions.
3/20/2026
Augmenting Rating-Scale Measures with Text-Derived Items Using the Information-Determined Scoring (IDS) Framework
Information-Determined Scoring framework using LLMs to score free-text psychological assessment responses and augment rating-scale measures.
3/20/2026
From Binary to Bilingual: How the National Weather Service is Using Artificial Intelligence to Develop a Comprehensive Translation Program
National Weather Service implementation of automated translation using LLMs and LILT's training process to serve non-English speakers.
3/20/2026
Manual2Skill++: Connector-Aware General Robotic Assembly from Instruction Manuals via Vision-Language Models
Robotic assembly system using vision-language models to handle connector-aware assembly from instruction manuals with focus on physical constraints.
3/20/2026
Open-o3-Video: Grounded Video Reasoning with Explicit Spatio-Temporal Evidence
Video reasoning model with explicit spatio-temporal evidence grounding, extending evidence-centered reasoning from images to videos with temporal tracking and spatial localization.
3/20/2026
The Geometry of Dialogue: Graphing Language Models to Reveal Synergistic Teams for Multi-Agent Collaboration
Framework for composing synergistic multi-agent LLM teams by analyzing model interaction geometry to optimize collaboration and surpass single-model capabilities.
3/20/2026
iSeal: Encrypted Fingerprinting for Reliable LLM Ownership Verification
Method for LLM ownership verification using encrypted fingerprinting with protection against attacks during verification processes.
3/20/2026
Seeing Beyond the Image: ECG and Anatomical Knowledge-Guided Myocardial Scar Segmentation from Late Gadolinium-Enhanced Images
Multimodal framework combining ECG signals and anatomical knowledge for cardiac myocardial scar segmentation from MRI images.
3/20/2026
Membership Inference Attack against Large Language Model-based Recommendation Systems: A New Distillation-based Paradigm
Novel knowledge distillation-based membership inference attack against LLM-based recommendation systems to determine if data samples were in training sets.
3/20/2026
Evolved Sample Weights for Bias Mitigation: Effectiveness Depends on the Fairness Objective
Study comparing genetic algorithms and other methods for generating sample weights to mitigate bias in ML models, examining trade-offs between fairness and accuracy.
3/20/2026
Steering Awareness: Detecting Activation Steering from Within
Research on models' ability to detect activation steering vectors injected into their residual streams during forward passes, revealing steering awareness in instruction-tuned models.
3/20/2026
MRD: Multi-resolution Retrieval-Detection Fusion for High-Resolution Image Understanding
MRD fusion approach for high-resolution image understanding in MLLMs combining retrieval-augmented generation with detection to prevent object fragmentation and false positives.
3/20/2026
Cell-cell Communication Inference and Analysis: Biological Mechanisms, Computational Approaches, and Future Opportunities
Survey of cell-cell communication inference from single-cell omics data, covering biological mechanisms and computational approaches for ligand-receptor interaction analysis.
3/20/2026
Heads collapse, features stay: Why Replay needs big buffers
Continual learning study revealing asymmetry in experience replay between feature-level and classifier-level forgetting, showing minimal buffers preserve representations but not predictions.
3/20/2026
ClinicalTrialsHub: Bridging Registries and Literature for Comprehensive Clinical Trial Access
ClinicalTrialsHub platform consolidating ClinicalTrials.gov with PubMed data extraction, increasing structured trial data access by 83.8% for patients and clinicians.
3/20/2026
A Multicenter Benchmark of Multiple Instance Learning Models for Lymphoma Subtyping from HE-stained Whole Slide Images
Benchmark of multiple instance learning models for lymphoma subtyping from whole slide images, comparing deep learning approaches for pathology diagnosis.
3/20/2026
Adaptive Accountability in Networked MAS: Tracing and Mitigating Emergent Norms at Scale
Adaptive Accountability Framework for networked multi-agent systems using cryptographic provenance tracking and runtime detection of emergent norms like collusion and unfairness.
3/20/2026
Neuron-Guided Interpretation of Code LLMs: Where, Why, and How?
Neuron-level interpretability study of code LLMs identifying language-specific neurons and concept layers, adapting NLP techniques to formal programming language structure.
3/20/2026
GeoMotionGPT: Geometry-Aligned Motion Understanding with Large Language Models
GeoMotionGPT aligns motion space geometry with embedding space in LLM-based motion understanding by coupling discrete motion tokenization with semantic learning.
3/20/2026
Vulnerability of LLMs' Stated Beliefs? LLMs Belief Resistance Check Through Strategic Persuasive Conversation Interventions
Systematic evaluation of LLM susceptibility to persuasion across six models using SMCR communication framework, testing adoption of counterfactual beliefs.
3/20/2026
Forest-Chat: Adapting Vision-Language Agents for Interactive Forest Change Analysis
Forest-Chat integrates vision-language agents with satellite imagery for interactive forest change analysis, combining LLMs with computer vision for environmental monitoring.
3/20/2026
Mechanism Shift During Post-training from Autoregressive to Masked Diffusion Language Models
Mechanistic study comparing internal algorithmic changes when post-training autoregressive models into masked diffusion models, investigating genuine bidirectional reasoning acquisition.
3/20/2026
The Flexibility Trap: Why Arbitrary Order Limits Reasoning Potential in Diffusion Language Models
Analysis of diffusion language models showing arbitrary token generation order doesn't unlock reasoning improvements over autoregressive models, revealing limitations of flexibility.
3/20/2026
STELLAR: Structure-guided LLM Assertion Retrieval and Generation for Formal Verification
STELLAR framework guides LLM-based generation of SystemVerilog Assertions for formal verification using structural similarity from hardware design ASTs.
3/20/2026
1S-DAug: One-Shot Data Augmentation for Robust Few-Shot Generalization
One-shot data augmentation method combining geometric perturbations with noise injection for few-shot learning generalization to novel classes.
3/20/2026
Sheaf Neural Networks and biomedical applications
Sheaf Neural Networks algorithm with biomedical case study outperforming GCNs, GATs, and GraphSage on graph-structured biomedical data.
3/20/2026
Gender Dynamics and Homophily in a Social Network of LLM Agents
Analysis of gender dynamics and homophily patterns in Chirper.ai, a social network of 70K+ autonomous LLM agents generating 140M posts, examining how AI agent identity develops in networks.
3/20/2026
Optimal rates for density and mode estimation with expand-and-sparsify representations
Theoretical study of expand-and-sparsify sparse representations for density and mode estimation, analyzing biological sensory system models with random projections and sparsification.
3/20/2026
Krause Synchronization Transformers
Krause Attention proposes a new transformer attention mechanism addressing representation collapse and attention sink phenomena through bounded normalization inspired by Krause dynamics.
3/20/2026
SF-RAG: Structure-Fidelity Retrieval-Augmented Generation for Academic Question Answering
SF-RAG improves retrieval-augmented generation for academic QA by preserving hierarchical document structure instead of flattening papers into chunks, enabling better evidence allocation under token constraints.
3/20/2026
Stable Deep Reinforcement Learning via Isotropic Gaussian Representations
Deep reinforcement learning stability improvement using isotropic Gaussian representations to handle non-stationary training dynamics.
3/20/2026
CeRA: Breaking the Linear Ceiling of Low-Rank Adaptation via Manifold Expansion
Parameter-efficient fine-tuning method using manifold expansion to overcome linear limitations of LoRA in complex reasoning tasks.
3/20/2026
Optimizer-Induced Low-Dimensional Drift and Transverse Dynamics in Transformer Training
Analysis of transformer training dynamics under AdamW optimizer identifying low-dimensional stable drift patterns in parameter evolution.
3/20/2026
Transformers Remember First, Forget Last: Dual-Process Interference in LLMs
Cognitive psychology-based study showing LLMs exhibit proactive interference dominance, with early information overriding recent conflicting context.
3/20/2026
Theory of Code Space: Do Code Agents Understand Software Architecture?
Benchmark evaluating whether code agents can understand multi-file software architecture through codebase exploration under partial observability.
3/20/2026
Farther the Shift, Sparser the Representation: Analyzing OOD Mechanisms in LLMs
Analysis of LLM internal representations showing increased sparsity with task difficulty and out-of-distribution shift across contexts.
3/20/2026
Deep Expert Injection for Anchoring Retinal VLMs with Domain-Specific Knowledge
Domain-specific enhancement of vision-language models for ophthalmic diagnosis by injecting expert knowledge to address perception and reasoning gaps.
3/20/2026
Adversarial Latent-State Training for Robust Policies in Partially Observable Domains
Reinforcement learning robustness method using adversarial latent-state training for partially observable environments.
3/20/2026
A Unified View of Drifting and Score-Based Models
Theoretical analysis connecting drifting models and score-based models through kernel-induced mean-shift discrepancy.
3/20/2026
Interleaving Scheduling and Motion Planning with Incremental Learning of Symbolic Space-Time Motion Abstractions
Task and motion planning approach combining scheduling with incremental learning for warehouse automation under resource and motion constraints.
3/20/2026
Thousand-GPU Large-Scale Training and Optimization Recipe for AI-Native Cloud Embodied Intelligence Infrastructure
Large-scale distributed training infrastructure for embodied AI using thousand GPUs and LeRobot framework with optimization recipes.
3/20/2026
WebWeaver: Breaking Topology Confidentiality in LLM Multi-Agent Systems with Stealthy Context-Based Inference
Security vulnerability analysis of LLM multi-agent systems showing inference attacks can extract communication topology without administrative access.
3/20/2026
Representation Finetuning for Continual Learning
Parameter-efficient fine-tuning method using representation finetuning for continual learning on pre-trained models with explicit optimization dynamics.
3/20/2026
A Simple Efficiency Incremental Learning Framework via Vision-Language Model with Nonlinear Multi-Adapters
Incremental learning framework using vision-language models with multi-adapter fine-tuning to improve efficiency and reduce memory requirements.
3/20/2026
Affect Decoding in Phonated and Silent Speech Production from Surface EMG
Study on decoding emotional affect from surface EMG during speech production using machine learning.