InfoCoBuild

6.7960 Deep Learning

6.7960 - Deep Learning (Fall 2024, MIT OCW). Instructors: Prof. Phillip Isola, Prof. Sara Beery, and Dr. Jeremy Bernstein. This course covers the fundamentals of deep learning, including both theory and applications. Topics include neural net architectures (MLPs, CNNs, RNNs, graph nets, transformers), geometry and invariances in deep learning, backpropagation and automatic differentiation, learning theory and generalization in high dimensions, and applications to computer vision, natural language processing, and robotics. (from ocw.mit.edu)

Lecture 03 - Approximation Theory

How well can you approximate a given function by a DNN? We will explore various facets of this issue, from universal approximation to Barron's theorem. And does increasing the depth probably help for expressivity?


Go to the Course Home or watch other lectures:

Lecture 01 - Introduction to Deep Learning
Lecture 02 - How to Train a Neural Net
Lecture 03 - Approximation Theory
Lecture 04 - Architectures: Grids
Lecture 05 - Architectures: Graphs
Lecture 06 - Generalization Theory
Lecture 07 - Scaling Rules for Optimization
Lecture 08 - Architectures: Transformers
Lecture 09 - Hacker's Guide to Deep Learning
Lecture 10 - Architectures: Memory
Lecture 11 - Representation Learning: Reconstruction-Based
Lecture 12 - Representation Learning: Similarity-Based
Lecture 13 - Representation Learning: Theory
Lecture 14 - Generative Models: Basics
Lecture 15 - Generative Models: Representation Learning Meets Generative Modeling
Lecture 16 - Generative Models: Conditional Models
Lecture 17 - Generization: Out-of-Distribution (OOD)
Lecture 18 - Transfer Learning: Models
Lecture 19 - Transfer Learning: Data
Lecture 20 - Scaling Laws
Lecture 21 - Language Models
Lecture 22 - NULL
Lecture 23 - Metrized Deep Learning
Lecture 24 - Interference Models for Deep Learning
Lecture 25 - Pytorch Tutorial