2/18/2026
Stabilizing Test-Time Adaptation of High-Dimensional Simulation Surrogates via D-Optimal Statistics
Test-time adaptation method for ML surrogates handling distribution shifts in high-dimensional simulations using D-optimal statistics.
2/18/2026
CrispEdit: Low-Curvature Projections for Scalable Non-Destructive LLM Editing
CrispEdit: scalable second-order LLM editing algorithm preserving capabilities while modifying targeted behavior without reward hacking.
2/18/2026
Operationalising the Superficial Alignment Hypothesis via Task Complexity
Proposes task complexity metric to operationalize superficial alignment hypothesis in LLMs, examining knowledge acquisition during pre-training vs post-training.
2/18/2026
TokaMind: A Multi-Modal Transformer Foundation Model for Tokamak Plasma Dynamics
Open-source multi-modal transformer foundation model for tokamak plasma dynamics supporting time-series, profiles, and video data.
2/18/2026
Universal priors: solving empirical Bayes via Bayesian inference and pretraining
Theoretical justification for using pretrained transformers on synthetic data to solve empirical Bayes problems via Bayesian inference.
2/18/2026
Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
Method for detecting annotation errors in video datasets using loss trajectory analysis to identify mislabeling and temporal ordering issues.
2/18/2026
Exploiting Layer-Specific Vulnerabilities to Backdoor Attack in Federated Learning
Analysis of backdoor attack vulnerabilities in federated learning systems, targeting layer-specific weaknesses in distributed model training.
2/18/2026
Weight space Detection of Backdoors in LoRA Adapters
Detection method for backdoor attacks in LoRA adapters by analyzing weight space without requiring test inputs or known triggers.
2/18/2026
Decision Making under Imperfect Recall: Algorithms and Benchmarks
Benchmark suite and algorithms for decision-making problems where agents have imperfect recall of previous information.
2/18/2026
When Remembering and Planning are Worth it: Navigating under Change
Study of memory and planning mechanisms for AI agents navigating dynamic environments with uncertain sensing and non-stationary conditions.
2/18/2026
Accelerating Large-Scale Dataset Distillation via Exploration-Exploitation Optimization
Optimization approach for dataset distillation that compresses large datasets into compact synthetic versions while preserving model performance.
2/18/2026
Unforgeable Watermarks for Language Models via Robust Signatures
Novel watermarking scheme for LLMs to verify content provenance and prevent false attribution of generated text.
2/18/2026
SCENE OTA-FD: Self-Centering Noncoherent Estimator for Over-the-Air Federated Distillation
SCENE proposes pilot-free aggregation for federated learning using over-the-air transmission with self-centering energy estimation.
2/18/2026
GMAIL: Generative Modality Alignment for generated Image Learning
GMAIL framework aligns generative model outputs with real images to prevent mode collapse when using synthetic data for training.
2/18/2026
FlashMem: Supporting Modern DNN Workloads on Mobile with GPU Memory Hierarchy Optimizations
FlashMem optimizes DNN inference on mobile GPUs by improving memory hierarchy for weight loading and execution strategies.
2/18/2026
The Vision Wormhole: Latent-Space Communication in Heterogeneous Multi-Agent Systems
Research on latent-space communication for multi-agent LLM systems to reduce overhead and information loss vs discrete text communication.
2/18/2026
GaiaFlow: Semantic-Guided Diffusion Tuning for Carbon-Frugal Search
Semantic-guided diffusion tuning for energy-efficient neural information retrieval and search ranking.
2/18/2026
TAROT: Test-driven and Capability-adaptive Curriculum Reinforcement Fine-tuning for Code Generation with Large Language Models
Test-driven reinforcement fine-tuning approach for improving code generation quality and robustness in LLMs.
2/18/2026
Molecular Design beyond Training Data with Novel Extended Objective Functionals of Generative AI Models Driven by Quantum Annealing Computer
Framework for optimizing molecular generation using deep learning integrated with quantum annealing computers.
2/18/2026
ExpertWeaver: Unlocking the Inherent MoE in Dense LLMs with GLU Activation Patterns
Method to convert dense pretrained LLMs into efficient Mixture-of-Experts models using GLU activation patterns.
2/18/2026
Quantifying construct validity in large language model evaluations
Analysis of construct validity in LLM benchmarks, addressing test contamination and annotator error in capability measurement.
2/18/2026
Latency-aware Human-in-the-Loop Reinforcement Learning for Semantic Communications
Reinforcement learning framework for semantic communications in radio networks with human feedback and latency constraints.
2/18/2026
Recursive Concept Evolution for Compositional Reasoning in Large Language Models
Recursive Concept Evolution method improves compositional reasoning in LLMs by evolving abstract representations beyond fixed latent spaces.
2/18/2026
Spanning the Visual Analogy Space with a Weight Basis of LoRAs
Visual analogy learning using LoRA basis weights to manipulate images through demonstration without text descriptions.
2/18/2026
Beyond Labels: Information-Efficient Human-in-the-Loop Learning using Ranking and Selection Queries
Human-in-the-loop learning framework using ranking and selection queries to extract richer information than binary labels.
2/18/2026
Enabling Low-Latency Machine learning on Radiation-Hard FPGAs with hls4ml
hls4ml framework for deploying fast, radiation-hard ML on FPGAs for high-energy physics calorimeter applications.
2/18/2026
Neural Scaling Laws for Boosted Jet Tagging
Scaling laws for neural networks in high-energy physics jet tagging, exploring compute scaling similar to LLM approaches.
2/18/2026
Perceptive Humanoid Parkour: Chaining Dynamic Human Skills via Motion Matching
Humanoid robot performing dynamic parkour through motion matching and perception-driven decision-making with long-horizon skill composition.
2/18/2026
Robust Deep Reinforcement Learning against Adversarial Behavior Manipulation
Study of behavior-targeted adversarial attacks on reinforcement learning and defenses using imitation learning without white-box access.
2/18/2026
Topological quantification of ambiguity in semantic search
Topological analysis of semantic ambiguity in sentence embeddings using persistent homology metrics for semantic search.
2/18/2026
Policy Gradients for Cumulative Prospect Theory in Reinforcement Learning
Policy gradient theorem for Cumulative Prospect Theory objectives in finite-horizon RL, generalizing standard policy gradients.
2/18/2026
ETGL-DDPG: A Deep Deterministic Policy Gradient Algorithm for Sparse Reward Continuous Control
DDPG algorithm with epsilon-t-greedy exploration for sparse reward reinforcement learning with polynomial sample complexity bounds.
2/18/2026
Just KIDDIN: Knowledge Infusion and Distillation for Detection of INdecent Memes
Framework combining knowledge distillation from LVLMs and knowledge graphs for detecting toxicity in memes.
2/18/2026
NeuroLifting: Neural Inference on Markov Random Fields at Scale
Graph neural network technique for scalable inference in large Markov Random Fields.
2/18/2026
Should You Use Your Large Language Model to Explore or Exploit?
Systematic evaluation of LLMs' exploration-exploitation tradeoff capabilities in contextual bandit tasks.
2/18/2026
Qronos: Correcting the Past by Shaping the Future... in Post-Training Quantization
Qronos post-training quantization algorithm that corrects weight and activation quantization errors iteratively.
2/18/2026
Latent Veracity Inference for Identifying Errors in Stepwise Reasoning
Veracity Search algorithm identifies errors in chain-of-thought reasoning steps in language models.
2/18/2026
Enhanced Generative Model Evaluation with Clipped Density and Coverage
Metrics for evaluating generative model quality using clipped density and coverage with calibration improvements.
2/18/2026
Deep Ignorance: Filtering Pretraining Data Builds Tamper-Resistant Safeguards into Open-Weight LLMs
Data filtering techniques to build safety safeguards into open-weight LLMs resistant to tampering attacks.
2/18/2026
Morephy-Net: An Evolutionary Multi-objective Optimization for Replica-Exchange-based Physics-informed Neural Operator Learning Networks
Morephy-Net uses multi-objective evolutionary optimization for physics-informed neural operators on parametric PDEs in noisy regimes.
2/18/2026
GenFacts-Generative Counterfactual Explanations for Multi-Variate Time Series
GenFacts generates valid counterfactual explanations for multivariate time series using class-discriminative VAE.
2/18/2026
Learning Admissible Heuristics for A*: Theory and Practice
Learns admissible heuristics for A* search algorithms using constrained optimization to guarantee solution optimality.
2/18/2026
Flock: A Knowledge Graph Foundation Model via Learning on Random Walks
Flock: knowledge graph foundation model using random walk learning for zero-shot link prediction on novel entities and relations.
2/18/2026
TabImpute: Universal Zero-Shot Imputation for Tabular Data
TabImpute: zero-shot universal imputation for tabular data with missing values using language model approach.
2/18/2026
General Exploratory Bonus for Optimistic Exploration in RLHF
General exploratory bonus method for optimistic exploration in RLHF that avoids bias toward reference model high-probability regions.
2/18/2026
Learning Pseudorandom Numbers with Transformers: Permuted Congruential Generators, Curricula, and Interpretability
Transformer models learn permuted congruential generator sequences via in-context prediction with curriculum learning and interpretability analysis.
2/18/2026
Terminal Velocity Matching
Terminal Velocity Matching generalizes flow matching for one/few-step generative modeling with Wasserstein distance bounds.
2/18/2026
Syndrome-Flow Consistency Model Achieves One-step Denoising Error Correction Codes
Applies consistency models to error correction codes for one-step neural decoding in low-latency communication settings.
2/18/2026
BEP: A Binary Error Propagation Algorithm for Binary Neural Networks Training
BEP algorithm trains binary neural networks with constrained weights/activations via error propagation for resource-constrained deployment.
2/18/2026
Improving Variational Autoencoder using Random Fourier Transformation: An Aviation Safety Anomaly Detection Case-Study
Improves VAE and autoencoder training using random Fourier transformation with frequency principle analysis for aviation safety anomaly detection.