Lecture 2A: Convolutional Neural Networks (Full Stack Deep Learning - Spring 2021)
In this video, we first review convolution operation, the most basic property of Convolutional Neural Networks. Then, we look at other important operations for ConvNets. Finally, we transition to looking at a classic ConvNet architecture called LeNet.
00:00 - Introduction
01:08 - Convolutional Filters
07:10 - Filter Stacks and ConvNets
11:25 - Strides and Padding
14:35 - Filter Math
21:44 - Convolution Implementation Notes
24:04 - Increasing the Receptive Field with Dilated Convolutions
27:30 - Decreasing the Tensor Size with Pooling and 1x1-Convolutions
30:54 - LeNet Architecture