ADAMS - L9 - Convolutional Neural Networks - Part 1

The last lecture of ADAMS introduces convolutional neural networks (CNNs). We first cover some fundamentals on how computers store and process images. This is to help us understand why feedforward NNs are not a suitable model for image data (because they lack scalability). After briefly introducing some common applications in computer vision, the core session elaborates on CNNs for image data. Thereafter, we touch on CNNs for non-image data and discuss how they can process, for example, text.