efore we begin this chapter, we need to talk a bit about AI and machine learning (ML) and how those two components fit together. The term "artificial" refers to something that is not real or natural, whereas "intelligence" refers to something capable of understanding, learning, or able to solve problems (and, in extreme cases, being self-aware).
Officially, artificial intelligence research began at the Dartmouth Conference of 1956 where AI and its mission were defined. In the following years, everyone was optimistic as machines were able to solve algebra problems and learn English, and the first robot was constructed in 1972. However in the 1970s, due to overpromising but under delivering, there was a so-called AI winter where AI research was limited and underfunded. After this though AI was reborn through expert systems, that could display human-level analytical skills. Afterwards, a second AI winter machine learning got recognized as a separate field in the 1990s when probability theories and statistics started to be utilized.
Increases in computational power and the determination to solve specific problems led to the development of IBM’s Deep Blue that beat the world chess champion in 1997 . Fast forward and nowadays the AI landscape encompasses many fields including Machine Learning, Computer Vision, Natural Language Processing, Planning Scheduling, and Optimization, Reasoning/Expert systems, and Robotics.
During the past 10 years, we have witnessed a huge transformation in what ML, and AI in general, is capable of. Thanks mainly to Deep Learning.
In this chapter, we are going to cover the following topics:
- A general explanation of the concepts of AI and ML
- Artificial neural networks and Deep Learning
- Convolutional neural networks (CNNs) and their main building blocks
- Using TensorFlow to build a CNN model to recognize images of digits
- An introduction to Tensorboard