2/16/2026
Designing RNAs with Language Models
Using language models for RNA sequence design by reframing it as conditional sequence generation task.
2/16/2026
Favia: Forensic Agent for Vulnerability-fix Identification and Analysis
LLM-based agent for identifying vulnerability-fixing commits in CVE security research with improved precision-recall.
2/16/2026
Bench-MFG: A Benchmark Suite for Learning in Stationary Mean Field Games
Benchmark suite for evaluating RL algorithms in stationary mean field games with standardized evaluation protocols.
2/16/2026
Decoder-only Conformer with Modality-aware Sparse Mixtures of Experts for ASR
Decoder-only Conformer with modality-aware sparse MoE for automatic speech recognition without external encoders.
2/16/2026
Monte Carlo Tree Search with Reasoning Path Refinement for Small Language Models in Conversational Text-to-NoSQL
MCTS-based approach for conversational text-to-NoSQL queries using small language models with reasoning path refinement.
2/16/2026
VI-CuRL: Stabilizing Verifier-Independent RL Reasoning via Confidence-Guided Variance Reduction
Verifier-free RL method for improving LLM reasoning without external verifiers, using confidence-guided variance reduction.
2/16/2026
QuEPT: Quantized Elastic Precision Transformers with One-Shot Calibration for Multi-Bit Switching
QuEPT: elastic precision quantization for Transformers/LLMs with one-shot calibration enabling multi-bit deployment scenarios.
2/16/2026
Self-EvolveRec: Self-Evolving Recommender Systems with LLM-based Directional Feedback
Self-EvolveRec: LLM-based recommender system evolution using directional feedback beyond scalar metrics, enabling open-ended program space search.
2/16/2026
Vision Token Reduction via Attention-Driven Self-Compression for Efficient Multimodal Large Language Models
Attention-driven token reduction for MLLMs via self-compression, pruning vision tokens through FlashAttention-compatible methods for efficiency.
2/16/2026
TensorCommitments: A Lightweight Verifiable Inference for Language Models
TensorCommitments: cryptographic verification protocol for LLM inference allowing clients to verify remote execution correctness without rerunning models.
2/16/2026
Unleashing Low-Bit Inference on Ascend NPUs: A Comprehensive Evaluation of HiFloat Formats
Evaluation of HiFloat low-bit formats (HiF8/HiF4) for LLM inference on Ascend NPUs, comparing with INT8 and other formats for efficiency.
2/16/2026
Artic: AI-oriented Real-time Communication for MLLM Video Assistant
Artic: real-time communication framework for multimodal LLM video assistants optimizing QoE and latency in cloud deployments.
2/16/2026
Beyond Normalization: Rethinking the Partition Function as a Difficulty Scheduler for RLVR
VI-RLVR: reinterprets partition function as difficulty scheduler for reward-maximizing LLM reasoning, improving output diversity while maintaining performance.
2/16/2026
Multi-Task Learning with Additive U-Net for Image Denoising and Classification
AddUNet: U-Net architecture with additive skip fusion for multi-task image denoising and classification with structural regularization.
2/16/2026
PMG: Parameterized Motion Generator for Human-like Locomotion Control
PMG: parameterized motion generator for humanoid locomotion control combining RL and motion tracking with adaptive interfaces.
2/16/2026
IndicFairFace: Balanced Indian Face Dataset for Auditing and Mitigating Geographical Bias in Vision-Language Models
IndicFairFace dataset for auditing geographical bias in vision-language models, addressing underrepresentation of Indian demographics with intra-national diversity.
2/16/2026
SLA2: Sparse-Linear Attention with Learnable Routing and QAT
SLA2: sparse-linear attention mechanism with learnable routing for accelerating diffusion and video generation models, includes quantization analysis.
2/16/2026
Trust the uncertain teacher: distilling dark knowledge via calibrated uncertainty
Knowledge distillation technique using calibrated uncertainty to transfer dark knowledge from teacher to student models while preserving probabilistic patterns.
2/16/2026
ALOE: Action-Level Off-Policy Evaluation for Vision-Language-Action Model Post-Training
ALOE: off-policy evaluation method for vision-language-action models using RL in real-world settings with value functions from mixed trajectory data sources.
2/16/2026
MedXIAOHE: A Comprehensive Recipe for Building Medical MLLMs
Medical vision-language foundation model achieving state-of-the-art performance on diverse medical benchmarks with entity-aware continual pretraining.
2/16/2026
"Not Human, Funnier": How Machine Identity Shapes Humor Perception in Online AI Stand-up Comedy
Study on how machine identity in AI chatbots affects humor perception in stand-up comedy, comparing to human comedians.
2/16/2026
Can Neural Networks Provide Latent Embeddings for Telemetry-Aware Greedy Routing?
Message Passing Network algorithm for explainable telemetry-aware routing in computer networks using latent embeddings.
2/16/2026
RAT-Bench: A Comprehensive Benchmark for Text Anonymization
Comprehensive benchmark for evaluating text anonymization tools used with LLMs, assessing effectiveness at preventing re-identification beyond basic PII removal.
2/16/2026
Left-right asymmetry in predicting brain activity from LLMs' representations emerges with their formal linguistic competence
Study examining how linguistic competence in LLMs correlates with left-right brain asymmetry when predicting human neural activity from text.
2/16/2026
GRAIL: Geometry-Aware Retrieval-Augmented Inference with LLMs over Hyperbolic Representations of Patient Trajectories
Framework combining retrieval-augmented inference with hyperbolic geometry representations of patient trajectories for clinical event prediction using LLMs.
2/16/2026
FLAC: Maximum Entropy RL via Kinetic Energy Regularized Bridge Matching
Likelihood-free reinforcement learning framework using kinetic energy regularization for iterative generative policies like diffusion models.
2/16/2026
TRACE: Temporal Reasoning via Agentic Context Evolution for Streaming Electronic Health Records (EHRs)
Agentic context evolution framework for LLMs to improve temporal reasoning over streaming electronic health records without fine-tuning or retrieval augmentation.
2/16/2026
Amortized Reasoning Tree Search: Decoupling Proposal and Decision in Large Language Models
Method for decoupling proposal and decision in LLM reasoning to address suppression of valid rare reasoning paths in reinforcement learning with verifiable rewards.
2/16/2026
Chimera: Neuro-Symbolic Attention Primitives for Trustworthy Dataplane Intelligence
Framework combining neural attention mechanisms with symbolic constraints for trustworthy, auditable inference on programmable network dataplanes.
2/16/2026
Knowledge-Based Design Requirements for Generative Social Robots in Higher Education
Framework defining knowledge prerequisites for Generative Social Robots in education, addressing hallucinations, overreliance, and privacy risks in LLM-based tutoring.
2/16/2026
RADAR: Revealing Asymmetric Development of Abilities in MLLM Pre-training
Framework for diagnosing performance bottlenecks in Multi-modal Large Language Models through efficient evaluation of asymmetric ability development.
2/16/2026
Robustness of Object Detection of Autonomous Vehicles in Adverse Weather Conditions
Method for evaluating robustness of object detection models in autonomous vehicles under adverse weather using synthetic data augmentation.
2/16/2026
Never say never: Exploring the effects of available knowledge on agent persuasiveness in controlled physiotherapy motivation dialogues
Research on how available knowledge affects persuasiveness of Generative Social Agents in physiotherapy dialogues, examining risks of manipulation in agent outputs.
2/16/2026
Transporting Task Vectors across Different Architectures without Training
Training-free method for transferring task-specific parameter updates across models with different architectures using learned transport functions.
2/16/2026
TriGen: NPU Architecture for End-to-End Acceleration of Large Language Models based on SW-HW Co-Design
NPU hardware architecture designed for efficient end-to-end LLM inference on resource-constrained devices through software-hardware co-design.
2/16/2026
RGAlign-Rec: Ranking-Guided Alignment for Latent Query Reasoning in Recommendation Systems
Recommendation system method using ranking-guided alignment to bridge semantic gaps between user features and LLM-based intent prediction for e-commerce.
2/16/2026
Drift-Aware Variational Autoencoder-based Anomaly Detection with Two-level Ensembling
VAE-based anomaly detection method with two-level ensembling for streaming data that handles concept drift in nonstationary environments.
2/16/2026
Learning Native Continuation for Action Chunking Flow Policies
Training method for action-chunked vision-language models that ensures smooth trajectory continuity at chunk boundaries via native continuation.
2/16/2026
Know More, Know Clearer: A Meta-Cognitive Framework for Knowledge Augmentation in Large Language Models
Meta-cognitive framework for knowledge augmentation in LLMs that addresses knowledge-confidence gaps and reduces overconfident errors.
2/16/2026
Synaptic Activation and Dual Liquid Dynamics for Interpretable Bio-Inspired Models
Bio-inspired neural network framework using chemical synapses and synaptic activation for improved interpretability in recurrent models.
2/16/2026
Prior-Guided Symbolic Regression: Towards Scientific Consistency in Equation Discovery
Symbolic regression method incorporating scientific priors to discover interpretable equations consistent with fundamental principles.
2/16/2026
Buy versus Build an LLM: A Decision Framework for Governments
Decision framework for governments evaluating whether to build, buy, or hybrid-adopt large language models for public sector applications.
2/16/2026
Look Inward to Explore Outward: Learning Temperature Policy from LLM Internal States via Hierarchical RL
Hierarchical reinforcement learning method to learn adaptive temperature policies from LLM internal states for improved sampling during training.
2/16/2026
Geometric Manifold Rectification for Imbalanced Learning
Geometric approach to imbalanced classification that rectifies class manifolds by addressing topological intrusion of majority class.
2/16/2026
Curriculum-DPO++: Direct Preference Optimization via Data and Model Curricula for Text-to-Image Generation
Curriculum-based training approach for Direct Preference Optimization in text-to-image generation, organizing learning by difficulty.
2/16/2026
Diverging Flows: Detecting Extrapolations in Conditional Generation
Technique for detecting when flow matching models produce implausible outputs for out-of-distribution conditions in safety-critical applications.
2/16/2026
Which Algorithms Can Graph Neural Networks Learn?
Analysis of graph neural networks' capacity to learn discrete algorithms, examining message-passing MPNNs for neural algorithmic reasoning integration.
2/16/2026
SCOPE: Selective Conformal Optimized Pairwise LLM Judging
SCOPE framework for LLM-based pairwise evaluation provides finite-sample statistical guarantees and calibration against miscalibration and systematic biases.
2/16/2026
In-Context Autonomous Network Incident Response: An End-to-End Large Language Model Agent Approach
LLM agent system for autonomous network incident response learns from system logs and alerts without handcrafted simulator modeling.
2/16/2026
Asynchronous Verified Semantic Caching for Tiered LLM Architectures
Asynchronous semantic caching framework for tiered LLM architectures reduces inference cost and latency in production agentic workflows.