A Toy Model for Hierarchical Feature Learning
Abstract
In this talk I will describe a toy model of deep learning for inputs with multiscale structure, in which features are extracted at each scale via a single-index model. I will describe connections with existing templates, and an initial analysis of statistically efficient learning algorithms.
Fields of Study & Subjects
Neural Networks and Applications
Stochastic Gradient Optimization Techniques
Machine Learning and Algorithms
Publication details
- Journal / source
- Open Collections
- Publication date
- 1 January 2026
- Language
- EN