Jul 18, 2025 · 8 min read · Why do we need layer normalization? Layer normalization is crucial in deep learning for several key reasons: Addressing Internal Covariate Shift: Stabilizes input distributions to each layer as previous layer parameters update during training Faster...
Join discussion
Jan 7, 2025 · 17 min read · In my last blog post, we discussed and solved through code, how to initialize the hidden layer in neural network properly, to get rid of dead neurons. We came up with some magic numbers, randomly, as well as, by looking at the mathematics of tanh act...
Join discussion
Oct 13, 2024 · 7 min read · Imagine you're in a group project with 10 people, and each person has to contribute an idea (data) to make the final presentation (prediction). But there's a problem: Some people talk too much, giving too many ideas. Some people are too quiet, bare...
Join discussionSep 28, 2024 · 11 min read · Normalization is one of the key ingredients of many fundamental principles that underpin effective neural network models in deep learning, a fast-growing domain. The normalization techniques are very crucial from input data up to hidden layer activat...
Join discussionFeb 7, 2024 · 3 min read · Introduction: In the ever-evolving landscape of deep learning, techniques like Batch Normalization have emerged as essential tools for enhancing the stability and efficiency of neural networks. In this blog post, we embark on a journey into the realm...
Join discussion
Oct 19, 2023 · 4 min read · In the ever-evolving realm of neural networks, ensuring that models generalize well to new, unseen data is a cardinal pillar. One of the key techniques that have emerged as a savior in enhancing model generalization is Batch Normalization. This power...
Join discussion
Sep 23, 2023 · 2 min read · Batch normalization has become a key technique for training deep neural networks. But what exactly does it do and why is it so important? In this post, we’ll explain batch normalization using an analogy, walk through the technical details, and cover ...
Join discussion