Papers

Cards List

An Exploratory Replica-Overlap Probe of the Grokking Transition

arXiv cs.LG · 16h ago Cached

This paper investigates the grokking transition in neural networks using replica-overlap probes, but reports challenges with the probe's validity and offers post-hoc statistical analysis.

0 favorites 0 likes

What Should a Self-Teacher See? Privileged Context Design for On-Policy Self-Distillation

arXiv cs.LG · 16h ago Cached

The paper investigates privileged context design in on-policy self-distillation, demonstrating that intermediate levels of abstraction can improve model performance over full solutions while using fewer hint tokens.

0 favorites 0 likes

Rewired or Gated? How Instruction Tuning Shapes Knowledge-Conflict Circuits in LLMs

arXiv cs.LG · 16h ago Cached

This paper provides a mechanistic comparison of knowledge-conflict circuits in LLMs under instruction tuning, finding that tuning gates rather than rewires these circuits across multiple model families, with implications for interpretability.

0 favorites 0 likes

EMGBlend: Heterogeneity-Aware Self-Supervised Pretraining for Gesture and Force Decoding

arXiv cs.LG · 16h ago Cached

EMGBlend introduces a self-supervised framework for pretraining on heterogeneous EMG datasets, addressing differences in electrode layouts, frequency support, and data source imbalances to improve gesture recognition and force decoding tasks.

0 favorites 0 likes

SambaGraph: Action-Reaction Spatio-Temporal Graphs for Soccer Tactical Response Modeling

arXiv cs.LG · 16h ago Cached

SambaGraph introduces a spatio-temporal graph dataset and benchmark for modeling soccer tactical responses, curated from 2022 FIFA World Cup data to classify actions and retrieve defensive examples.

0 favorites 0 likes

DefaultGNN: A Dual-Perspective GNN Framework for Predicting Corporate Default from Buyer-Seller Transaction Networks

arXiv cs.LG · 16h ago Cached

DefaultGNN is a dual-perspective graph neural network framework for predicting corporate default using buyer-seller transaction networks, achieving improvements over baselines and validating with real-world data.

0 favorites 0 likes

A JEPA Recipe for Tabular Foundation Models

arXiv cs.LG · 16h ago Cached

This paper proposes a training recipe for tabular foundation models using a Joint-Embedding Predictive Architecture (JEPA) to address latent term collapse, and compares its performance to a value-only approach across 147 real datasets.

0 favorites 0 likes

Hill Sampling for Test-Time Scaling: A Simple and Better Alternative to Repeated Sampling, Evolution, and Training

arXiv cs.LG · 16h ago Cached

Hill Sampling is a simple test-time scaling method that repeatedly samples edits to the best verified program using frozen LLMs, achieving state-of-the-art results on algorithmic problems like circle packing and Erdős' minimum-overlap problem.

0 favorites 0 likes

Continuous Optimization for p-adic Models

arXiv cs.LG · 16h ago Cached

This paper introduces the first method for continuous gradient descent optimization in machine learning models with p-adic parameters, using the Berkovich affine line to enable effective learning on tasks like modular arithmetic.

0 favorites 0 likes

Learning Defensive Policies against Diverse Inference Attacks for Smart Meter Privacy

arXiv cs.LG · 16h ago Cached

This paper proposes a proxy-guided hierarchical reinforcement learning framework to defend against diverse inference attacks on smart meter data by learning battery-based load-shaping policies that disrupt non-intrusive load monitoring patterns.

0 favorites 0 likes

Terminal Shrinkage Averaging Reveals a Schedule-Estimator Interaction in LLM Pretraining

arXiv cs.LG · 16h ago Cached

Proposes Terminal Shrinkage Averaging (TSA) to separate learning-rate schedule from model estimator in LLM pretraining, improving validation quality and potentially accelerating benchmarks.

0 favorites 0 likes

A Practical Recipe for Semi-Supervised Federated ASR: Online Pseudo-Labels with Server Update Stabilization

arXiv cs.LG · 16h ago Cached

This paper proposes a practical recipe for semi-supervised federated ASR using online pseudo-labels with server update stabilization, demonstrating significant improvements over prior methods in both in-domain and cross-domain settings.

0 favorites 0 likes

Mean Velocity Matching: Rethinking Generative Dynamics in Diffusion Models

arXiv cs.LG · 16h ago Cached

This paper introduces Mean Velocity Matching (MVM) to parameterize stochastic reverse dynamics in diffusion models using a single learned field, enabling both stochastic and deterministic sampling with competitive generation quality.

0 favorites 0 likes

PermuFormer: Multi-Task Pretraining for Permutation Representation in Algebraic Combinatorics

arXiv cs.LG · 16h ago Cached

This paper introduces PermuFormer, an autoregressive transformer pretrained on multi-task, multi-encoding data for permutation-focused tasks in algebraic combinatorics, demonstrating effective fine-tuning on downstream tasks compared to baselines.

0 favorites 0 likes

Lightweight Ranking Heads: Accelerating Multi-Task Experimentation in Production Recommender Systems

arXiv cs.LG · 16h ago Cached

This paper introduces the Lightweight Ranking Heads framework to accelerate multi-task experimentation in production recommender systems by enabling dynamic task injection without retraining backbone models, reducing iteration cycles from weeks to days at YouTube scale.

0 favorites 0 likes

Predictive Uncertainty for Neural CAE Surrogates

arXiv cs.LG · 16h ago Cached

The paper investigates uncertainty quantification methods for neural surrogates in computer-aided engineering, comparing Gaussian processes, Monte Carlo dropout, and deep ensembles on industrial datasets to evaluate their performance in various engineering contexts.

0 favorites 0 likes

Deep Reinforcement Learning on Item-Compatibility Graphs for One-Dimensional Bin Packing

arXiv cs.LG · 16h ago Cached

This paper introduces a graph-based deep reinforcement learning framework for the one-dimensional bin packing problem, reducing optimality gaps compared to existing methods and enabling zero-shot generalization across instance sizes.

0 favorites 0 likes

Extending FunctionGemma for Practical On-Device Mobile Function Calling

arXiv cs.LG · 16h ago Cached

This research paper extends FunctionGemma 270M for practical on-device Android workflows by introducing a synthetic dataset and fine-tuning the model, achieving improved accuracy for function calling while balancing performance and coverage.

0 favorites 0 likes

Concept Drift from a Causal Perspective

arXiv cs.LG · 16h ago Cached

This paper proposes a causal framework for understanding concept drift in data streams using Structural Causal Models, with a taxonomy and generator for simulating and evaluating drift events in non-stationary environments.

0 favorites 0 likes

MT-ProtBERT: Multi-task Learning ProtBERT for Intrinsically Disordered Proteins Classification with Scarce Data

arXiv cs.LG · 16h ago Cached

MT-ProtBERT is a multi-task learning model for classifying intrinsically disordered proteins under data scarcity, integrating self-supervised and biochemistry-informed tasks to outperform existing methods like PARROT.

0 favorites 0 likes
← Previous
Next →
← Back to home

Submit Feedback