2/16/2026
Synaptic Activation and Dual Liquid Dynamics for Interpretable Bio-Inspired Models
Framework analyzing bio-inspired RNN models with chemical synapses and synaptic activation to improve interpretability of recurrent neural networks.
2/16/2026
FedHENet: A Frugal Federated Learning Framework for Heterogeneous Environments
FedHENet extends federated learning to image classification using fixed feature extractors, reducing privacy risks and computational costs in heterogeneous environments.
2/16/2026
Curriculum-DPO++: Direct Preference Optimization via Data and Model Curricula for Text-to-Image Generation
Curriculum-DPO++ applies curriculum learning to direct preference optimization for improved text-to-image generation model training.
2/16/2026
AdaGrad-Diff: A New Version of the Adaptive Gradient Algorithm
AdaGrad variant using cumulative squared gradient differences instead of norms for adaptive stepsize learning.
2/16/2026
Compressible Dynamics in Deep Overparameterized Low-Rank Learning & Adaptation
Method leveraging low-rank structures and compressible parameter dynamics to reduce computational costs in overparameterized neural network learning.
2/16/2026
LTSM-Bundle: A Toolbox and Benchmark on Large Language Models for Time Series Forecasting
LTSM-Bundle: toolbox and benchmark for using large language models as universal time series forecasting models on heterogeneous datasets.
2/16/2026
Generating Physical Dynamics under Priors
Framework for incorporating physical priors into diffusion-based generative models to generate physically feasible dynamics.
2/16/2026
Explaining and Mitigating the Modality Gap in Contrastive Multimodal Learning
Analysis of the modality gap in contrastive multimodal learning (CLIP-like models) and methods to explain and reduce representation misalignment between modalities.
2/16/2026
B3C: A Minimalist Approach to Offline Multi-Agent Reinforcement Learning
B3C: minimalist offline multi-agent reinforcement learning approach using behavior cloning regularization to address overestimation in action selection.
2/16/2026
Memory Injection Attacks on LLM Agents via Query-Only Interaction
MINJA attack: adversarial method for injecting malicious data into LLM agent memory banks via query-only interactions without direct memory access.
2/16/2026
Learning on a Razor's Edge: Identifiability and Singularity of Polynomial Neural Networks
Theoretical analysis of identifiability and singularity in polynomial neural networks and their neuromanifolds.
2/16/2026
Optimal Formats for Weight Quantisation
Framework for systematic design and analysis of weight quantization formats for efficient deep learning model training and deployment.
2/16/2026
Beyond All-to-All: Causal-Aligned Transformer with Dynamic Structure Learning for Multivariate Time Series Forecasting
Transformer model with dynamic structure learning for multivariate time series forecasting using causal relationships instead of all-to-all connections.
2/16/2026
On Learning Verifiers and Implications to Chain-of-Thought Reasoning
Research on learning verifiers for chain-of-thought reasoning in LLMs using formal verification to improve mathematical problem-solving.
2/16/2026
N$^2$: A Unified Python Package and Test Bench for Nearest Neighbor-Based Matrix Completion
N², a unified Python package for nearest neighbor-based matrix completion with theoretical guarantees and empirical benchmarks.
2/16/2026
PeakWeather: MeteoSwiss Weather Station Measurements for Spatiotemporal Deep Learning
PeakWeather dataset of Swiss weather station measurements for training spatiotemporal deep learning models for weather forecasting.
2/16/2026
Holistic Continual Learning under Concept Drift with Adaptive Memory Realignment
Framework for continual learning that handles concept drift in real-world data streams while preventing catastrophic forgetting.
2/16/2026
Adopting a human developmental visual diet yields robust, shape-based AI vision
Research on improving AI vision robustness by training on human developmental visual patterns to reduce reliance on texture and increase shape recognition.
2/16/2026
FISHER: A Foundation Model for Multi-Modal Industrial Signal Comprehensive Representation
Foundation model for analyzing industrial signals from SCADA systems across multiple modalities for anomaly detection.
2/16/2026
Instruction-based Time Series Editing
Instruction-based diffusion model for editing time series properties while preserving specified conditions, replacing rigid attribute vectors.
2/16/2026
Bridging Generalization Gap of Heterogeneous Federated Clients Using Generative Models
Federated learning approach using generative models to address data heterogeneity across distributed clients and improve model generalization.
2/16/2026
R-Zero: Self-Evolving Reasoning LLM from Zero Data
Self-evolving LLM that autonomously generates training data and improves reasoning without relying on human-curated tasks or labels.
2/16/2026
Semantic Caching for Low-Cost LLM Serving: From Offline Learning to Online Adaptation
Semantic caching system for reducing LLM inference costs by retrieving cached responses based on query similarity rather than exact matches.
2/16/2026
MissionHD: Hyperdimensional Refinement of Distribution-Deficient Reasoning Graphs for Video Anomaly Detection
Hyperdimensional refinement method for LLM-generated reasoning graphs in video anomaly detection handling distribution-deficient structures.
2/16/2026
Multipole Semantic Attention: A Fast Approximation of Softmax Attention for Pretraining
Fast approximate softmax attention mechanism clustering queries and keys for 36% faster transformer pretraining on long sequences.
2/16/2026
Online reinforcement learning via sparse Gaussian mixture model Q-functions
Online reinforcement learning framework using sparse Gaussian mixture model Q-functions with interpretable policy iteration.
2/16/2026
Kairos: Toward Adaptive and Parameter-Efficient Time Series Foundation Models
Adaptive time series foundation model with parameter-efficient design handling temporal heterogeneity and varying sampling rates.
2/16/2026
VDW-GNNs: Vector diffusion wavelets for geometric graph neural networks
Vector diffusion wavelets integrated into geometric graph neural networks for point cloud and manifold data representation.
2/16/2026
Provable Training Data Identification for Large Language Models
Statistical set-level inference framework to identify training data in large language models with controlled error rates for legal evidence.
2/16/2026
Don't Walk the Line: Boundary Guidance for Filtered Generation
Reinforcement learning method to fine-tune generative models avoiding decision boundary mode-seeking when paired with safety classifiers.
2/16/2026
Weight Decay may matter more than muP for Learning Rate Transfer in Practice
Weight decay regularization has greater impact than maximal update parameterization for transferring learning rates across model scales.
2/16/2026
Active Learning with Task-Driven Representations for Messy Pools
Active learning approach using task-driven representations to select relevant samples from uncurated, messy data pools.
2/16/2026
LLMs as In-Context Meta-Learners for Model and Hyperparameter Selection
Large language models used as in-context meta-learners to recommend model families and hyperparameters from dataset metadata.
2/16/2026
Membership and Dataset Inference Attacks on Large Audio Generative Models
Methods to detect if specific audio data was used in training generative audio models through membership and dataset inference attacks.
2/16/2026
ATLAS: Adaptive Topology-based Learning at Scale for Homophilic and Heterophilic Graphs
Graph neural network framework for handling heterophilic graphs without iterative message passing using multi-resolution community features.
2/16/2026
Imitation Learning for Combinatorial Optimisation under Uncertainty
Imitation learning framework for combinatorial optimization under uncertainty. Explores role of expert demonstrations in training policies for sequential decision problems.
2/16/2026
Spectral Ghost in Representation Learning: from Component Analysis to Self-Supervised Learning
Study of spectral ghost phenomenon in self-supervised learning. Analyzes representation learning using unlabeled data for downstream task transfer.
2/16/2026
SAGE: Sequence-level Adaptive Gradient Evolution for Generative Recommendation
RL-based preference optimization for generative recommenders. Proposes SAGE to address symmetric conservatism failure in list-wise ranking with multi-objective feedback.
2/16/2026
Gauss-Newton Natural Gradient Descent for Shape Learning
Gauss-Newton natural gradient descent method for shape learning addressing ill-conditioning in implicit neural surfaces.
2/16/2026
Pareto-Conditioned Diffusion Models for Offline Multi-Objective Optimization
Pareto-Conditioned Diffusion framework for offline multi-objective optimization using conditional sampling.
2/16/2026
AEGIS: Adversarial Target-Guided Retention-Data-Free Robust Concept Erasure from Diffusion Models
Method for erasing harmful concepts from diffusion models while maintaining robustness and generation quality.
2/16/2026
tLoRA: Efficient Multi-LoRA Training with Elastic Shared Super-Models
tLoRA framework for efficient multi-LoRA training on frozen LLM backbones with elastic shared super-models.
2/16/2026
Learnable Chernoff Baselines for Inference-Time Alignment
Learnable Chernoff Baselines for efficient inference-time alignment of generative models using reward guidance.
2/16/2026
Thermodynamic Isomorphism of Transformers: A Lagrangian Approach to Attention Dynamics
Thermodynamic framework analyzing transformer attention through Lagrangian mechanics and entropy minimization.
2/16/2026
LLaDA2.1: Speeding Up Text Diffusion via Token Editing
LLaDA2.1 text diffusion improvement combining token-to-token and mask-to-token editing for faster generation.
2/16/2026
Epistemic Throughput: Fundamental Limits of Attention-Constrained Inference
Formalization of attention-constrained inference for screening and verifying candidates under limited review capacity.
2/16/2026
Low-Dimensional Execution Manifolds in Transformer Learning Dynamics: Evidence from Modular Arithmetic Tasks
Analysis showing transformer learning dynamics collapse onto low-dimensional manifolds despite high parameter dimensionality.
2/16/2026
Diffusion-Pretrained Dense and Contextual Embeddings
Multilingual embedding models using contrastive learning on diffusion-pretrained backbone for web-scale retrieval.
2/16/2026
HiFloat4 Format for Language Model Inference
HiFloat4 block floating-point format for efficient LLM inference, achieving 4.5 bits per value with three-level scaling.
2/16/2026
Online Tensor Inference
Online tensor inference method for real-time processing of sequentially arriving high-dimensional data with statistical capabilities.