Chapter 1. Toward deep learning: a machine-learning introduction
Figure 1.1. The standard programming paradigm that most software developers are familiar with. The developer identifies the algorithm and implements the code; the users supply the data.
Figure 1.2. The machine-learning paradigm: during development, you generate an algorithm from a data set, and then incorporate that into your final application.
Figure 1.3. A simple example data set. Each point on the graph represents a soccer player’s height and weight. Your goal is to fit a model to these points.
Figure 1.4. First you note that your data set roughly follows a linear trend, then you find the formula for a specific line that fits the data.
Figure 1.5. Machine-learning algorithms operate on mathematical structures, such as vectors and matrices. Your photo tags are stored in a standard computer data structure: a list of strings. This is one possible scheme for encoding that list as a mathematical vector.
Figure 1.6. A machine-learning pipeline for supervised learning
Figure 1.7. An unsupervised machine-learning pipeline for finding clusters or chunks of chess pieces
Figure 1.8. In reinforcement learning, agents learn to interact with their environment by trial and error. You repeatedly have your agent attempt its task to get a supervised signal to learn from. With every cycle, you can make an incremental improvement.
Figure 1.9. Deep learning and representation learning