Modular and Productive Deep Learning Code
Tirthajyoti Sarkar · Apress eBooks · 2022
In the previous chapter, I explored the idea that most data scientists often come from a background that is quite far removed from traditional computer science/software engineering. Consequently, they produce code that is perfectly suitable for great exploratory data analysis, statistical modeling, or innovative ML experiments, but not robust enough for the production phase of a large business platform. Data scientists often think in terms of the next analysis script but not along the lines of the next software module that integrates into a larger system. In this chapter, we will explore how similar principles of modularized coding can help you write better code for deep learning tasks with some hands-on examples using Keras/TensorFlow.