4/23/2026
MOMO: A framework for seamless physical, verbal, and graphical robot skill learning and adaptation
Interactive framework for robot skill adaptation using kinesthetic, natural language, and graphical modalities for non-expert users.
4/23/2026
Forecasting Individual NetFlows using a Predictive Masked Graph Autoencoder
Graph Neural Network model for predicting network traffic flow patterns by modeling graph structure and connection features in heterogeneous bidirectional graphs.
4/23/2026
Decentralized Machine Learning with Centralized Performance Guarantees via Gibbs Algorithms
Decentralized machine learning approach using Gibbs algorithms that achieves centralized performance without sharing local datasets across clients.
4/23/2026
Towards Certified Malware Detection: Provable Guarantees Against Evasion Attacks
Certifiably robust malware detection framework using randomized smoothing through feature ablation to defend against adversarial evasion attacks.
4/23/2026
Efficient Symbolic Computations for Identifying Causal Effects
Efficient symbolic computation methods for determining causal effect identifiability from observational data with latent confounding in linear structural causal models.
4/23/2026
On Bayesian Softmax-Gated Mixture-of-Experts Models
Theoretical study of Bayesian mixture-of-experts models with softmax gating mechanisms, exploring their probabilistic properties and learning dynamics.
4/23/2026
An explicit operator explains end-to-end computation in the modern neural networks used for sequence and language modeling
Research paper establishing mathematical correspondence between state space models and nonlinear oscillator networks, analyzing S4D sequence modeling architecture.
4/23/2026
Self-Aware Vector Embeddings for Retrieval-Augmented Generation: A Neuroscience-Inspired Framework for Temporal, Confidence-Weighted, and Relational Knowledge
Research paper proposing temporal, confidence-weighted, relational metadata for vector embeddings in RAG systems to improve accuracy on versioned queries.
4/23/2026
pAI/MSc: ML Theory Research with Humans on the Loop
pAI/MSc: Open-source multi-agent system for automating academic ML research workflows from hypothesis to manuscript draft with human guidance.
4/23/2026
Auto-ART: Structured Literature Synthesis and Automated Adversarial Robustness Testing
Auto-ART framework synthesizes adversarial robustness research across 2020-2026 and provides automated testing protocols to detect gradient masking.
4/23/2026
AAC: Admissible-by-Architecture Differentiable Landmark Compression for ALT
Introduces AAC, a differentiable landmark compression module for shortest-path heuristics with admissible-by-construction guarantees.
4/23/2026
V-tableR1: Process-Supervised Multimodal Table Reasoning with Critic-Guided Policy Optimization
V-tableR1 framework uses process-supervised RL to train multimodal LLMs for rigorous multi-step table reasoning with verifiable rewards.
4/23/2026
Coverage, Not Averages: Semantic Stratification for Trustworthy Retrieval Evaluation
Proposes semantic stratification method to improve retrieval evaluation for RAG systems by addressing bias in query set construction and metric reliability.
4/23/2026
Working Memory Constraints Scaffold Learning in Transformers under Data Scarcity
Integrates working memory constraints into transformers via attention variants, evaluates on language understanding tasks.
4/23/2026
Convergent Evolution: How Different Language Models Learn Similar Number Representations
Identifies convergent evolution of periodic number representations across different LLM architectures and training methods.
4/23/2026
Issues with Value-Based Multi-objective Reinforcement Learning: Value Function Interference and Overestimation Sensitivity
Analyzes value function interference and overestimation issues in multi-objective reinforcement learning algorithms.
4/23/2026
FlashNorm: Fast Normalization for Transformers
Optimizes RMSNorm computation in LLMs by eliminating normalization weights through matrix folding for parallel execution.
4/23/2026
Towards Certified Unlearning for Deep Neural Networks
Extends certified unlearning methods from convex models to deep neural networks with theoretical guarantees.
4/23/2026
Verification of Machine Unlearning is Fragile
Demonstrates vulnerabilities in machine unlearning verification strategies used to validate data removal from models.
4/23/2026
MixLLM: LLM Quantization with Global Mixed-precision between Output-features and Highly-efficient System Design
Proposes mixed-precision quantization for LLMs that varies precision across output features for efficient compression.
4/23/2026
Best Policy Learning from Trajectory Preference Feedback
Studies best policy identification in preference-based RL as alternative to learned reward models for alignment.
4/23/2026
Recency Biased Causal Attention for Time-series Forecasting
Introduces recency bias mechanism for transformer attention in time-series forecasting via reweighted attention scores.
4/23/2026
Not All Rollouts are Useful: Down-Sampling Rollouts in LLM Reinforcement Learning
Proposes PODS to decouple rollout generation from policy updates in LLM reinforcement learning with verifiable rewards.
4/23/2026
Kalman Filter Enhanced GRPO for Reinforcement Learning-Based Language Model Reasoning
Improves advantage estimation in language model RL by applying Kalman filtering to Group Relative Policy Optimization.
4/23/2026
On the definition and importance of interpretability in scientific machine learning
Examines interpretability definitions in scientific ML and barriers to integrating neural network findings into scientific knowledge.
4/23/2026
Latent Stochastic Interpolants
Extends stochastic interpolants framework to latent space with joint optimization of encoder-decoder for generative modeling.
4/23/2026
On the Existence of Universal Simulators of Attention
Theoretical analysis of transformer expressivity and learnability through the lens of universal simulators of attention mechanisms.
4/23/2026
Optimal Single-Policy Sample Complexity and Transient Coverage for Average-Reward Offline RL
Theoretical analysis of offline reinforcement learning in average-reward MDPs with sample complexity guarantees under single-policy data coverage.
4/23/2026
FeDa4Fair: Client-Level Federated Datasets for Fairness Evaluation
Dataset and evaluation methodology for assessing fairness in federated learning at client level, addressing discrimination across distributed clients.
4/23/2026
The Ratchet Effect in Silico through Interaction-Driven Cumulative Intelligence in Large Language Models
POLIS framework enabling heterogeneous LLM agents to accumulate knowledge through interaction and emergent communication, modeling cumulative cultural evolution.
4/23/2026
FLOSS: Federated Learning with Opt-Out and Straggler Support
Federated learning framework addressing privacy and stragglers by supporting user opt-out and handling missing data from heterogeneous devices.
4/23/2026
KANMixer: a minimal KAN-centered mixer for long-term time series forecasting
KANMixer architecture using Kolmogorov-Arnold Networks for long-term time series forecasting with improved expressivity over MLP and Transformer baselines.
4/23/2026
Local Diffusion Models and Phases of Data Distributions
Research on diffusion models that incorporate local spatial structure into score functions for better generative modeling of spatially structured data.
4/23/2026
From Nodes to Narratives: Explaining Graph Neural Networks with LLMs and Graph Context
Explanation method for Graph Neural Networks using LLMs to generate interpretable narratives from node-level decisions and graph context.
4/23/2026
WISCA: A Lightweight Model Transition Method to Improve LLM Training via Weight Scaling
WISCA improves LLM training by systematically optimizing weight patterns through weight scaling transitions during transformer training.
4/23/2026
Evaluating the Quality of the Quantified Uncertainty for (Re)Calibration of Data-Driven Regression Models
Evaluates calibration metrics and recalibration methods for uncertainty estimation in data-driven regression models for safety-critical applications.
4/23/2026
EvolveSignal: A Large Language Model Powered Coding Agent for Discovering Traffic Signal Control Strategies
EvolveSignal uses LLM-powered coding agent to automatically discover traffic signal control strategies, replacing manual hand-crafted formulas.
4/23/2026
Representational Alignment Across Model Layers and Brain Regions with Multi-Level Optimal Transport
Multi-Level Optimal Transport method for symmetric representational alignment across model layers and different network depths.
4/23/2026
Distributional Inverse Reinforcement Learning
Distributional framework for offline inverse RL capturing richer expert behavior by modeling uncertainty over rewards and return distributions.
4/23/2026
From Competition to Synergy: Unlocking Reinforcement Learning for Subject-Driven Image Generation
Applies reinforcement learning with GRPO to subject-driven image generation, addressing fidelity-editability trade-off via synergistic reward alignment.
4/23/2026
Understanding the Staged Dynamics of Transformers in Learning Latent Structure
Investigates how transformers learn latent structure through staged dynamics using controlled Alchemy benchmark tasks.
4/23/2026
Hybrid-AIRL: Enhancing Inverse Reinforcement Learning with Supervised Expert Guidance
Hybrid-AIRL combines adversarial inverse RL with supervised expert guidance, evaluated on poker domain with imperfect information.
4/23/2026
A Unified Theory of Sparse Dictionary Learning in Mechanistic Interpretability: Piecewise Biconvexity and Spurious Minima
Unified theoretical framework for sparse dictionary learning in mechanistic interpretability, analyzing piecewise biconvexity and spurious minima in neural networks.
4/23/2026
Bridging Mechanistic Interpretability and Prompt Engineering with Gradient Ascent for Interpretable Persona Control
Novel framework combining mechanistic interpretability with gradient ascent for interpretable persona control in LLMs, improving on unscalable prompt engineering.
4/23/2026
CEDAR: Context Engineering for Agentic Data Science
CEDAR demonstrates automated data science task solving using LLM agents via context engineering, addressing task complexity, data size, and computational constraints.
4/23/2026
Agnostic Language Identification and Generation
Theoretical analysis of language identification and generation without realizability assumptions, establishing statistical rates under relaxed conditions.
4/23/2026
SwiftRepertoire: Few-Shot Immune-Signature Synthesis via Dynamic Kernel Codes
Few-shot learning framework for T cell receptor repertoire analysis using dynamic kernel codes and prototype-based parameterizations for disease detection.
4/23/2026
Rashomon Sets and Model Multiplicity in Federated Learning
Analysis of Rashomon sets in federated learning to understand model multiplicity and decision boundary instabilities affecting transparency and robustness.
4/23/2026
Explainability in Generative Medical Diffusion Models: A Faithfulness-Based Analysis on MRI Synthesis
Faithfulness-based explainability framework for analyzing generative diffusion models in medical MRI synthesis applications.
4/23/2026
Towards Explainable Federated Learning: Understanding the Impact of Differential Privacy
Study analyzing explainability in federated learning systems with differential privacy layers, examining privacy-interpretability trade-offs.