Artificial Intelligence - what is it and how to get to know it better?
02 Oct 2019 | Machine Learning
Machine Learning is slowly becoming synonymous with Artificial Intelligence, and around the mysterious name of Data Science accumulated so many myths that it has lost its original meaning long ago. Personally, I consider myself more of a mathematician rather than a programmer because of my mathematical background, which in turn is not so rare among Data Scientists. It will not surprise anyone that I like to use terms that are ordered within a certain system. Therefore, let's have a look at the AI ecosystem together to introduce some order.
What is the hierarchy of individual AI terms?
The most general term is Artificial Intelligence. It’s the name of the entire field of science dedicated to creating a program that will autonomously solve all problems known to mankind. Science distinguishes many types of artificial intelligence and various methods of achieving it. There are many scenarios for reaching an autonomous thinking being. Interestingly, most of them assume that at some point an intelligent program will improve itself, which can cause a significant threat to the human species.
- Checkers - AI wins
- Backgammon - AI wins
- Othello - AI wins
- Chess - AI wins
- Crosswords - man wins, but AI is already at the expert level
- Scrabble - AI wins
- Bridge - man wins, but AI is already at the expert level
- Go - AI wins
AI has to face many challenges, such as computer vision, natural language processing, decision making or Big Data etc. Each of these challenges can be implemented in different ways, and some of them I will try to bring a little closer.
One of the most interesting and promising methods are artificial neural networks called deep learning. They are inspired by real connections of the neural network and, like the human brain, can learn to propagate signals in a manner corresponding to the desired functionality.
A slightly different philosophy is behind reinforcement learning - it is a group of algorithms that interact with the environment and get penalties or rewards depending on the activity performed. This causes the algorithm to act in the desired way.
Principles of machine learning
Theoretically, from a mathematical point of view, deep learning belongs to Machine Learning methods. However, due to a different philosophy in developing these methods, focused more on heuristics and Deep Learning applications, it is often treated as a separate field of science.
Anyway, let's go back to Machine Learning itself. These methods are based mainly on statistical properties which allow us to conclude from some dependencies found in data. We are lucky to live in a very interesting period for this science field. We may even be able to watch the birth of the first super-intelligent program. Personally, it fills me with fear rather than optimism. If you're more curious about dangers posed by artificial intelligence, let me refer you to the book called 'Superintelligence’ by Nick Bostrom.
Complete familiarization with the field of artificial intelligence takes years of studying the existing achievements of mankind in this area. However, some methods, mainly for business purposes, can easily be learned by participating in the bootcamp organized by fireup.pro. This is a 60-hour, complete machine learning course including the basics of AI and IoT (Internet of Things). We split it into two parts: the intro and hands-on experience, which will allow us to reliably and methodically discuss the most important aspects of the issues raised. We also focused on the practical aspect of the course - this means that for every class, in addition to acquiring theoretical knowledge, we will also do a workshop. The workshop will, for example, involve a simulation of a simple probabilistic system, preparing a model for classifying objects in photos or training a neural network to paint images or generate song lyrics.
More about the course can be found at: https://fireup.pro/machine-learning/
A curious and inquisitive man. He scrabbles in data for fun. Seriously interested in the development of artificial intelligence methods in science and business. He got into machine learning already in college, as a result of which he wrote a master's thesis in mathematics on the dynamics of chaotic neural networks.