Posted Date: 06/06/2022

Artificial intelligence series overview

On March 9, 2016, a shocking incident occurred in which Go player Sedol Lee lost against Google's artificial intelligence (AI) AiphaGo. People who don't usually play Go, those who have nothing to do with the IT industry, and those who think that artificial intelligence is a subject that only appears in movies and cartoons also paid attention to this. This process is reflected in the documentary.

AiphaGo was implemented by deep learning, which is the core technology of artificial intelligence, and was an opportunity to spark interest in deep learning in the general public after the match with Sedol Lee. On the other hand, machine learning, machine learning as an artificial intelligence technology has also begun to be recognized by the public, and the relationship between artificial intelligence, machine learning, and deep learning can be summarized as follows.

Machine learning belongs to a large category of artificial intelligence, and a part of machine learning is classified as deep learning. By analogy, artificial intelligence corresponds to general food, machine learning corresponds to more advanced food, and deep learning corresponds to high-end steak dishes. Therefore, the flower of artificial intelligence can be called deep learning. This does not mean, however, that other technologies in machine learning other than deep learning are useless. It's just the choice of which technology to apply to the situation is different. Naive Bayes algorithm, a type of machine learning technology, is typically applied to the technology of filtering spam in the mail space we encounter every day.

In this series, we want to focus on Bayesian inference underlying deep learning and artificial intelligence technologies. The basis of deep learning is an artificial neural network. Although the Bayesian Neural Network model, which has recently evolved an artificial neural network, is attracting attention, it focuses on artificial neural networks and Bayesian inference itself.

Everyone talks about artificial intelligence. My Facebook news feed is also covered with artificial intelligence content. Articles related to artificial intelligence account for a significant portion of newspapers, and electronic products need to include the term artificial intelligence to sell well. There is even a joke that if you enter the word artificial intelligence in your business plan, you can receive investment. Although this is likely to cause controversy over bubbles, artificial intelligence, especially deep learning, is proving itself day by day by putting out amazing results to the world. In addition, many artificial intelligence software libraries have emerged, becoming a world where even general software developers can easily use artificial intelligence. However, there was a time when artificial intelligence was reserved only for 'doctors'. This is because the theoretical structure of these artificial intelligence technologies is very mathematical and specialized, so it is theoretically overkill for the general public and general software developers.

However, if this series is for general developers, it will mainly be filled with contents about libraries and source code fragments such as Python, Anaconda, TensorFlow, and Keras. The main purpose is to explore the content itself of artificial intelligence rather than to explore what kind of story these technologies consist of and how they work. There will be some mathematical and technical content, but there will be no need to worry at all as it will be developed with common sense content so that ordinary business people can fully understand and access it.