Tech Xplore

Overparameterized neural networks: Feature learning precedes overfitting, research finds

Modern neural networks, with billions of parameters, are so overparameterized that they can "overfit" even random, ...

Anthropic vs. OpenAI red teaming methods reveal different security priorities for enterprise AI

Anthropic runs 200-attempt attack campaigns. OpenAI reports single-attempt metrics. A 16-dimension comparison reveals what ...
IEEE

Activating Sparse Part Concepts for 3D Class Incremental Learning

Abstract: This work tackles the challenge of 3D Class-Incremental Learning (CIL), where a model must learn to classify new 3D objects while retaining knowledge of previously learned classes. Existing ...
TMCnet

NTT Scientists Contribute Fifteen Research Papers to NeurIPS 2025

One of the three primary annual conferences in ML and AI research, NeurIPS 2025 is taking place Dec. 2-7 at the San Diego Convention Center.
IEEE

Curriculum Learning Algorithms for Reward Weighting in Sparse Reward Robotic Manipulation Tasks

Abstract: Robotic learning from sparse rewards can be a considerable challenge due to large amounts of data required for mastering a task. We explore the application of curriculum learning (CL) ...