Lilian Weng • 9/28/2017

Anatomize Deep Learning with Information Theory

Read Original

This article summarizes Professor Naftali Tishby's work on using information theory to study deep learning. It explains concepts like the Information Bottleneck method, which provides a new learning bound for DNNs, and describes the two-phase training process of fitting data and then compressing representation. Key information theory concepts such as Markov chains, KL divergence, mutual information, and the Data Processing Inequality are covered in the context of neural networks.

0 comments
#Neural Networks #Deep Learning #Information Theory
Anatomize Deep Learning with Information Theory

Comments

No comments yet

Be the first to share your thoughts!

Browser Extension

Get instant access to AllDevBlogs from your browser

Top of the Week

1
React, useEffect, and Stale Closures: How to Avoid Them with Dependencies and useRef
Jivbcoop • 4 votes
2
Using Browser Apis In React Practical Guide
Jivbcoop • 2 votes
3
Better react-hook-form Smart Form Components
Maarten Hus • 2 votes
4
Top picks — 2026 January
Paweł Grzybek • 1 votes
5
In Praise of –dry-run
Henrik Warne • 1 votes
6
Deep Learning is Powerful Because It Makes Hard Things Easy - Reflections 10 Years On
Ferenc Huszár • 1 votes
7
Vibe coding your first iOS app
William Denniss • 1 votes
8
AGI, ASI, A*I – Do we have all we need to get there?
John D. Cook • 1 votes
9
Quoting Thariq Shihipar
Simon Willison • 1 votes
10
Dew Drop – January 15, 2026 (#4583)
Alvin Ashcraft • 1 votes