Interesting AI/ML Articles You Should Read This Week (Aug 31)

An explanation of an integral component within a widely used convolutional neural network architecture. Andre Ye writes about Google's Inception architecture and the reason for its success in image classification and detection. Andre explains in this article that the Inception architecture success is primarily due to the efficient utilisation of the neural network architecture computing resources through the incorporation of 1x1 convolutions. This article explores in impressive detail as to why 1x1 convolutions are an integral component o the Inception CNN architecture. With the inclusion of detailed descriptions and creative illustrations, Andre is able to point out how convolutions operate on different formats of images (grayscale and RGB).