3/2/2026
The False Promise of Zero-Shot Super-Resolution in Machine-Learned Operators
Evaluation of zero-shot super-resolution capabilities in machine-learned operators for modeling continuous physical phenomena.
3/2/2026
Thompson Sampling via Fine-Tuning of LLMs
ToSFiT: Thompson sampling via LLM fine-tuning for Bayesian optimization in large discrete spaces without acquisition function maximization.
3/2/2026
Unified Privacy Guarantees for Decentralized Learning via Matrix Factorization
Differential privacy framework for decentralized learning using matrix factorization to enable collaborative training while preserving privacy.
3/2/2026
FAPO: Flawed-Aware Policy Optimization for Efficient and Reliable Reasoning
FAPO method for LLM reasoning via reinforcement learning that filters flawed positive rollouts to improve policy optimization with verifiable rewards.
3/2/2026
DiffuMamba: High-Throughput Diffusion LMs with Mamba Backbone
DiffuMamba: diffusion language model with Mamba backbone for efficient masked sequence modeling, achieving linear-time complexity vs Transformer quadratic overhead.
3/2/2026
Descend or Rewind? Stochastic Gradient Descent Unlearning
Analysis of stochastic gradient descent-based unlearning algorithms (D2D and R2D) with provable guarantees for removing training data impact.
3/2/2026
Heterogeneous Multi-Agent Reinforcement Learning with Attention for Cooperative and Scalable Feature Transformation
Multi-agent reinforcement learning approach using attention for automated feature transformation in structured data processing.
3/2/2026
Log Probability Tracking of LLM APIs
Method for monitoring LLM API consistency over time by tracking log probability changes to detect undisclosed model updates.
3/2/2026
On the Effectiveness of Membership Inference in Targeted Data Extraction from Large Language Models
Study on membership inference attacks for extracting training data from LLMs, demonstrating privacy risks through coordinated extraction and verification techniques.
3/2/2026
Smoothing DiLoCo with Primal Averaging for Faster Training of LLMs
Generalized Primal Averaging (GPA) optimizer for faster LLM training, extending Nesterov's method and unifying recent averaging-based approaches like DiLoCo.
3/2/2026
Trust Region Masking for Long-Horizon LLM Reinforcement Learning
Trust region masking technique for LLM reinforcement learning to address off-policy divergence in policy gradient training of large language models.
3/2/2026
CSyMR: Benchmarking Compositional Music Information Retrieval in Symbolic Music Reasoning
CSyMR benchmark for evaluating LLMs on compositional music information retrieval tasks requiring multi-step reasoning over symbolic music notation.
3/2/2026
DUET: Distilled LLM Unlearning from an Efficiently Contextualized Teacher
DUET method for LLM unlearning using distilled teacher models to remove undesirable knowledge efficiently while avoiding catastrophic forgetting.
3/2/2026
Federated-inspired Single-cell Batch Integration in Latent Space
Federated-inspired batch correction for single-cell RNA sequencing without centralizing high-dimensional datasets.
3/2/2026
Position: Beyond Model-Centric Prediction -- Agentic Time Series Forecasting
Position paper proposing agentic framework for time series forecasting with iterative refinement and adaptation.
3/2/2026
Quant VideoGen: Auto-Regressive Long Video Generation via 2-Bit KV-Cache Quantization
KV-cache quantization to 2-bit precision enabling long video generation on resource-constrained hardware.
3/2/2026
Erase at the Core: Representation Unlearning for Machine Unlearning
Machine unlearning method addressing superficial forgetting by targeting core feature representations rather than logits.
3/2/2026
Robust Online Learning
Online learning framework for training robust classifiers under adversarially chosen clean data and labels.
3/2/2026
Unified Biomolecular Trajectory Generation via Pretrained Variational Bridge
Pretrained variational bridge for efficient molecular dynamics trajectory generation across diverse molecular systems.
3/2/2026
Biases in the Blind Spot: Detecting What LLMs Fail to Mention
Automated detection pipeline for unverbalized biases in LLM chain-of-thought reasoning without predefined categories.
3/2/2026
RooflineBench: A Benchmarking Framework for On-Device LLMs via Roofline Analysis
Performance characterization framework for small language models on edge devices using Roofline model analysis.
3/2/2026
It's TIME: Towards the Next Generation of Time Series Forecasting Benchmarks
Benchmarking framework for time series foundation models addressing data quality, task alignment, and evaluation rigor.
3/2/2026
Capabilities Ain't All You Need: Measuring Propensities in AI
Framework for measuring propensities (behavioral tendencies) in AI models beyond capability assessment using Item Response Theory.
3/2/2026
Exponential Convergence of (Stochastic) Gradient Descent for Separable Logistic Regression
Theoretical analysis of gradient descent convergence rates under large step sizes in separable logistic regression.
3/2/2026
Discrete Diffusion with Sample-Efficient Estimators for Conditionals
Discrete diffusion framework using sample-efficient conditional probability estimators for discrete state space generation.
3/2/2026
Test-Time Training with KV Binding Is Secretly Linear Attention
Analysis showing test-time training with KV binding functions as learned linear attention rather than memorization.
3/2/2026
FedVG: Gradient-Guided Aggregation for Enhanced Federated Learning
Federated learning aggregation method (FedVG) using gradient guidance to address client drift and data heterogeneity.
3/2/2026
Provably Safe Generative Sampling with Constricting Barrier Functions
Safety filtering framework for flow-based generative models with formal guarantees on constraint satisfaction.
3/2/2026
Training Generalizable Collaborative Agents via Strategic Risk Aversion
Strategic risk aversion approach for training collaborative AI agents that generalize better with new partners.
3/2/2026
To Deceive is to Teach? Forging Perceptual Robustness via Adversarial Reinforcement Learning
Adversarial reinforcement learning dataset (AOT-SFT) to improve robustness of multimodal LLMs on visually complex scenes.
3/2/2026
Manifold of Failure: Behavioral Attraction Basins in Language Models
Maps failure regions in LLMs using MAP-Elites quality diversity search to characterize unsafe behaviors and vulnerabilities.
3/2/2026
When Should a Model Change Its Mind? An Energy-Based Theory and Regularizer for Concept Drift in Electrocardiogram (ECG) Signals
Energy-based theory for detecting concept drift in ECG signals, distinguishing physiologically plausible variation from true distribution shift.
3/2/2026
Regularized Online RLHF with Generalized Bilinear Preferences
Regularized online RLHF for Nash Equilibrium identification with generalized bilinear preferences modeling intransitive preferences.
3/2/2026
ParamMem: Augmenting Language Agents with Parametric Reflective Memory
ParamMem augments language agents with parametric reflective memory to improve reasoning through diverse self-reflection.
3/2/2026
FinBloom: Knowledge Grounding Large Language Model with Real-time Financial Data
FinBloom presents a knowledge-grounding approach for LLMs to handle real-time financial queries with live data integration.
3/2/2026
Apple: Toward General Active Perception via Reinforcement Learning
Apple proposes general active perception via reinforcement learning to handle uncertainty in partially observable robotic environments.
3/2/2026
REA-RL: Reflection-Aware Online Reinforcement Learning for Efficient Reasoning
REA-RL uses online reinforcement learning with reflection to reduce overthinking and inference costs in large reasoning models.
3/2/2026
CoMind: Towards Community-Driven Agents for Machine Learning Engineering
CoMind introduces MLE-Live, a framework for evaluating LLM agents that engage with research communities in ML engineering tasks.
3/2/2026
Knowledge-Guided Machine Learning: Illustrating the use of Explainable Boosting Machines to Identify Overshooting Tops in Satellite Imagery
Uses Explainable Boosting Machines to identify overshooting tops in satellite imagery for weather forecasting, emphasizing interpretability in high-stakes ML.
3/2/2026
pFedMMA: Personalized Federated Fine-Tuning with Multi-Modal Adapter for Vision-Language Models
pFedMMA proposes personalized federated fine-tuning with multi-modal adapters for vision-language models like CLIP on decentralized heterogeneous data.
3/2/2026
FeynTune: Large Language Models for High-Energy Theory
20 fine-tuned Llama-3.1 8B variants specialized for High-Energy Physics, trained on arXiv abstracts with comparative domain analysis.
3/2/2026
Alignment through Meta-Weighted Online Sampling: Bridging the Gap between Data Generation and Preference Optimization
Research method for LLM alignment using meta-weighted online sampling to address distribution mismatch between offline data and evolving model policy.
3/2/2026
Scaling Generalist Data-Analytic Agents
DataMind framework trains scalable generalist data-analytic agents on diverse data formats for multi-step reasoning in scientific discovery.
3/2/2026
Unraveling Syntax: How Language Models Learn Context-Free Grammars
Study analyzing how language models learn context-free grammar structures and subgrammars during pretraining.
3/2/2026
Embracing Discrete Search: A Reasonable Approach to Causal Structure Learning
FLOP algorithm performs causal structure discovery for linear models using fast parent selection and discrete search optimization.
3/2/2026
Supervised Reinforcement Learning: From Expert Trajectories to Step-wise Reasoning
Supervised Reinforcement Learning framework combines expert trajectories with step-wise reasoning to improve multi-step reasoning in open-source LLMs.
3/2/2026
Steering Language Models with Weight Arithmetic
Contrastive weight steering method edits LLM parameters via weight arithmetic for post-training behavior modification without expensive retraining.
3/2/2026
Structure tensor Reynolds-averaged Navier-Stokes turbulence models with equivariant neural networks
Equivariant neural networks improve RANS turbulence model closures using structure tensors for better generalization in fluid dynamics.
3/2/2026
NuBench: An Open Benchmark for Deep Learning-Based Event Reconstruction in Neutrino Telescopes
NuBench open benchmark for deep learning-based event reconstruction in neutrino telescopes addressing inverse problems in particle physics.
3/2/2026
Score-Regularized Joint Sampling with Importance Weights for Flow Matching
Non-IID sampling framework for flow matching models to reduce variance in expectation estimation with limited sampling budgets.