Recurrent Neural Networks (RNN)

Channel:
Singapore John Tan Chong Min
Subscribers:
5,330
Published on ● Video Link: https://www.youtube.com/watch?v=SLHplbMGUKY


Duration: 1:18:14
87 views
1

An introduction to RNNs. Would also like to share that at 27:08, although I highlighted a random initial state (typically small noise to prevent overfitting), the TensorFlow default implementation is a zero initial state. It is also possible to learn the initial state using Backpropagation Through Time (BPTT).

Also, at 42:24, I used the CLS token to denote end of sequence. If we were to use the BERT architecture, it should be SEP to avoid confusion.

Slides and Code can be found at: https://github.com/tanchongmin/TensorFlow-Implementations



Other Videos By John Tan Chong Min


2022-06-08Geometric Deep Learning (Part 2)
2022-06-06Transformers (Part 2)
2022-06-01Equation Solver!
2022-05-30Transformers (Part 1)
2022-05-30Can a Neural Network model if-else statements?
2022-05-29Minesweeper AI using Neural Network!
2022-05-29Minesweeper AI using Neural Network! (Part 2)
2022-05-28CodinGame : Let's try to solve the Minesweeper Puzzle! (Part 2)
2022-05-28CodinGame: Let's try to solve the Minesweeper Puzzle!
2022-05-15CodinGame: Pushing rank (again) in Clash of Code!
2022-05-09Recurrent Neural Networks (RNN)
2022-05-01CodinGame: Spring Challenge - Gold League Attempt!
2022-05-01CodinGame: Spring Challenge - Can I make it to Gold?
2022-04-27CodinGame: Spring Challenge - Monte Carlo Wild Mana Collection
2022-04-26CodinGame: Spring Challenge - Let's build a game lookahead simulation!
2022-04-25Convolutional Neural Networks (Part 2)
2022-04-21CodinGame: Spring Challenge 2022 First Look
2022-04-21Can a Neural Network decode my message? (Part 3)
2022-04-21Can a Neural Network decode my message? (Part 2)
2022-04-20Can a Neural Network decode my message?
2022-04-18Convolutional Neural Networks (Part 1)