Implementing a Deep Learning Library from Scratch in Python - KDnuggets

Deep Learning has evolved from simple neural networks to quite complex architectures in a short span of time. To support this rapid expansion, many different deep learning platforms and libraries are developed along the way. One of the primary goals for these libraries is to provide easy to use interfaces for building and training deep learning models, that would allow users to focus more on the tasks at hand. To achieve this, it may require to hide core implementation units behind several abstraction layers that make it difficult to understand basic underlying principles on which deep learning libraries are based. Hence the goal of this article is to provide insights on building blocks of deep learning library.