Back Propagation Through Time (BPTT) in RNN | Vanishing Gradient | Exploding Gradient | Explained
About this lesson
๐ Notes: https://robosathi.com/docs/natural_language_processing/rnn/#bptt ๐ฅ NLP Course: https://www.youtube.com/playlist?list=PLnpa6KP2ZQxcDlHCeNiKbRhLWKVunQaxn ๐ฅ Deep Learning Course: https://www.youtube.com/playlist?list=PLnpa6KP2ZQxe749nPGDV2cd6SR6zIZIJl ๐ฅ Machine Learning Course: https://www.youtube.com/playlist?list=PLnpa6KP2ZQxeydAqz2lsSMFYinbrJy9mu ๐ฅ Full Maths Course: https://www.youtube.com/playlist?list=PLnpa6KP2ZQxen-R6NytSMigAri7piPhFp ๐ฅ Back Propagation: https://youtu.be/iDQTFJCOkHk ๐ฅ RNN: https://youtu.be/jSj1STepMA8 โ This video describes in detail with Maths how the gradient is Back Propagated Through Time in a RNN to update the weights. โ We will also understand the reason for Vanishing/Exploding gradient problem and solutions for them. ๐ Time Stamp ๐ 00:00:00 - 00:01:56 Introduction 00:01:57 - 00:04:11 RNN Architecture 00:04:12 - 00:05:37 RNN Parameters 00:05:38 - 00:12:20 What is Back Propagation Through Time (BPTT) ? 00:12:21 - 00:17:22 RNN Layers, Error & Gradient (Maths) 00:17:23 - 00:20:25 BPTT Gradient Calculation 00:20:26 - 00:24:15 Vanishing/Exploding Gradient Problem Example 00:24:16 - 00:28:11 Solution: Vanishing/Exploding Gradient Problem 00:28:12 - 00:29:16 Next: Applications of RNN
DeepCamp AI