Hands-On Machine Learning for Algorithmic Trading

Reference ConvNet architectures

Several CNN architectures have pushed performance boundaries over the past two decades by introducing important innovations. Predictive performance growth accelerated dramatically with the arrival of big data in the form of ImageNet, with 14 million images assigned to 20,000 classes by humans using Amazon MTurk. The ILSVRC became the focal point of CNN progress around a slightly smaller set of 1.2 million images from 1,000 classes.

It is useful to be familiar with the reference architectures resulting from these competitions for practical reasons. As we will see in the next section, Transfer learning – faster training with less data, many computer vision tasks can be addressed by building on a successful architecture ...