Convolutional Neural Networks (CNNs) | Deep Learning

CNNs are a go-to deep learning architecture for many computer vision tasks, from image classification to object detection and more. Here, we take a look at the basics, and see how they use biologically-inspired hierarchical feature extraction to do what they do. Timestamps -------------------- Introduction 00:00 Kernel convolutions 00:41 Common kernels 02:30 Why flipping? 03:30 Convolution as feature extraction 04:00 Hierarchical feature extraction 05:40 Down-sizing 07:25 Max-pooling 09:43 Multi-channel kernels 10:30 Learnable kernels 11:04 CNN architecture 11:52 Residual connections 14:25 Convolution vs. cross-correlation 15:55 Useful links ------------------- ResNet paper: https://arxiv.org/abs/1512.03385 CNN tutorial in TensorFlow: https://www.tensorflow.org/tutorials/images/cnn Gradient calculation in convolutional layers: https://deeplearning.cs.cmu.edu/F21/document/recitation/Recitation5/CNN_Backprop_Recitation_5_F21.pdf and https://sites.cc.gatech.edu/classes/AY2021/cs7643_spring/assets/L11_CNNs.pdf