Despite all the recent buzz about deep learning, the design and testing of a neural network pipeline may become a task for developers who aren't machine learning specialists. This tutorial is intended for a software developer who has intermediate experience in Python, plus some hands-on experience developing data pipelines and working with machine learning use cases, who now needs to learn how to build high-performance classifiers based on deep learning.
The Keras library in Python makes building and testing neural networks a snap. It provides a simpler, quicker alternative to Theano or TensorFlow–without worrying about floating point operations, GPU programming, linear algebra, etc. The Keras API should seem familiar for anyone who's worked with the well-known and well-loved scikit-learn API. This tutorial begins by building a simple classifier in a couple of lines of scikit-learn, then compares how to accomplish the same thing with a neural network using Keras. From there it moves into tougher classification problems involving image recognition that help illustrate the power of deep learning.
Here's a clip (or watch the full tutorial on O'Reilly's Learning Platform):
A few months ago, I started a new job doing applied research at Fast Forward Labs. My new colleagues had just finished a project on image recognition with convolutional neural networks, and were starting on a new project on text summarization with recurrent neural networks. Neural networks were not something I'd previously worked with, so I knew I'd have to get up to speed fast. Keras is a high-level neural network library that, among many other things, wraps an API similar to scikit-learn's around the Theano or TensorFlow backends.