Practical Java Machine Learning: Projects with Google Cloud Platform and Amazon Web Services by Mark Wickham

Author:Mark Wickham [Mark Wickham]
Language: eng
Format: epub, pdf
Publisher: Apress
Published: 2018-10-22T21:00:00+00:00

Neural networks have many types of hidden layers. Refer to the link to distinguish the different hidden layer types because it is hard to visualize the layer types in the greyscale image.

In the bottom left corner of Figure 4-7 is the support vector machine DL algorithm. The SVM DL algorithm is a supervised ML algorithm you can also apply to CML. You will take a closer look at the performance of this algorithm later in this chapter and again in Chapter 5.

Reinforcement Learning

Semi-supervised learning is sometimes confused with the reinforcement learning (RL) style . They are not the same. RL is a type of supervised learning with a distinction. With RL, each input does not always generate feedback. While semi-supervised learning uses data with mixed labels, with RL, there are no labels.

In RL, the supervision comes from a reward signal that tells the critic how well it is doing, but does not say what the correct action should be. Reinforcement learning deals with the interaction of the critic with its environment (state). The actions taken by the critic influence the distribution of states it will observe in the future. In supervised learning, each decision is independent of the others. In RL, the labels are associated with sequences, as opposed to individual samples in supervised learning.

Recall from Chapter 1, the Pokerbot problem was difficult to solve because poker is a game of uncertain or incomplete information. RL works well for navigating uncertain environments, and is thus often used for games such as poker, chess, blackjack, or Go.

Earlier, I mentioned Skymind, the creator of the Java-based DL library. Skymind also has some great content on RL. It described RL as a goal-oriented approach to ML. You can learn more about RL from the following link:

https://skymind.ai/wiki/deep-reinforcement-learning

In the rest of this chapter and book, you will restrict your focus to supervised and unsupervised algorithms with or without labels because they overlap well with your focus on CML problems.

Download

Copyright Disclaimer:
This site does not store any files on its server. We only index and link to content provided by other sites. Please contact the content providers to delete copyright contents if any and email us, we'll remove relevant links or contents immediately.