Skip to content
lectern.tv
MIT 6.7960: Deep Learning →0/20
1

Introduction to Deep Learning

1:00:51
2

How to Train a Neural Net

1:19:34
3

Approximation Theory

1:22:42
4

Architectures: Grids

1:23:57
5

Architectures: Graphs

1:21:14
6

Generalization Theory

1:20:31
7

Scaling Rules for Optimization

1:20:56
8

Architectures: Transformers

1:14:35
9

Hacker's Guide to Deep Learning

1:15:51
10

Architectures: Memory

1:13:28
11

Representation Learning: Reconstruction-Based

1:21:04
12

Representation Learning: Similarity-Based

1:16:20
13

Representation Learning: Theory

1:15:21
14

Generative Models: Basics

1:21:18
15

Generative Models: Representation Learning Meets Generative Modeling

1:20:40
16

Generative Models: Conditional Models

1:21:32
17

Generalization: Out-of-Distribution (OOD)

1:04:41
18

Transfer Learning: Models

1:25:42
19

Transfer Learning: Data

1:15:44
20

Scaling Laws

38:23

How to Train a Neural Net

Sara BeeryMIT

Professor

1:19:34·2026·graduate level

This session explains neural network training via stochastic gradient descent (SGD), backpropagation, and automatic differentiation. It highlights their role in differentiable programming.

AI & MLMathematicsDeep LearningProbability & Information TheoryConvex OptimizationNeural Network Training

MIT 6.7960 Deep Learning, Fall 2024

Instructor: Sara Beery

View the complete course: https://ocw.mit.edu/courses/6-7960-deep-learning-fall-2024/

YouTube Playlist: https://www.youtube.com/playlist?list=PLUl4u3cNGP63URZnh5iqBzDTDYPUTQT-8

Previous · Lecture 1

Introduction to Deep Learning

Next · Lecture 3

Approximation Theory

Comments

Add a comment...Sign in