Published inTowards Data Science·Nov 27, 2018Residual blocks — Building blocks of ResNetUnderstanding a residual block is quite easy. In traditional neural networks, each layer feeds into the next layer. In a network with residual blocks, each layer feeds into the next layer and directly into the layers about 2–3 hops away. That’s it. But understanding the intuition behind why it was…Machine Learning6 min readMachine Learning6 min read
Published inTowards Data Science·Nov 19, 2018Separable convolutions — trading little accuracy for huge computational gainsTypically in convolutions, we use a 2D or a 3D kernel filter where we hope that each filter extracts some kind of a feature by convoluting in all the 2 or 3 dimensions, respectively. Specifically in 2D case, we try to extract simple features in initial layer and more complex…Machine Learning3 min readMachine Learning3 min read
Published inTowards Data Science·Nov 18, 2018Grouped Convolutions — convolutions in parallelUsually, convolution filters are applied on an image layer by layer to get the final output feature maps. We increase the no of kernels per layer to learn more no of intermediate features, therefore increasing the no of channels in the next layer. But to learn more no. of features…Machine Learning4 min readMachine Learning4 min read
Published inTowards Data Science·Aug 19, 2018Deciding optimal kernel size for CNNConvolutional Neural Networks (CNNs) are neural networks that automatically extract useful features (without manual hand-tuning) from data-points like images to solve some given task like image classification or object detection. And now that you understand their use on your datasets, you start wondering: Apart from tuning various hyper-parameters of your…Machine Learning3 min readMachine Learning3 min read