Getting started in the field of Artificial Intelligence

Updated 29 Oct

Welcome! You’re about to embark on an awesome journey.

Generally, I believe the following skills to be important when diving into a Machine Learning / AI career or hobby, although this is clearly not an exclusive list:

• A Mathematical background: Probability Theory, Calculus, Linear Algebra
• A programming background: Python (become familiar with libraries like numpy, matplotlib), eventually you will dive into Deep Learning frameworks like TensorFlow or PyTorch (which one is best? I don’t know)
• AI “basics”: Machine Learning, Deep Learning
• Applied AI: Natural Language Processing, Computer Vision, Reinforcement Learning
• AI Safety: How do we keep systems safe?

Books

• Life 3.0

  A gentle, easy-to-read, introduction to the importance of the (correct) development of Artificial Intelligence for the future of humanity.

Study books

This section will link to some of the study books used in the Artificial Intelligence Master’s Program of the University of Amsterdam. Most are the classic books of their respective fields, and are highly recommended for anyone seriously going into the field.

• Reinforcement Learning, an Introduction

  Richard S. Sutton and Andrew G. Barto

• Pattern Recognition and Machine Learning

  Christopher M. Bishop

• Deep Learning

  Ian Goodfellow, Yoshua Bengio, Aaron Courville

Online Lecture Series

• Machine Learning by Stanford Professor Andrew Ng.

  There are also recorded ‘real’ lectures, and more recent ones. See which format you like best. Prerequisites: Some calculus, linear algebra.

• Reinforcement Learning by DeepMind Professor David Silver

  The RL course everyone always refers to. Prerequisites: Machine Learning, Deep Learning

Blog Posts

• Natural Language Processing

  • Recurrent Neural Networks
  • • The Unreasonable Effectiveness of Recurrent Neural Networks](http://karpathy.github.io/2015/05/21/rnn-effectiveness/)

Kaggle

Kaggle is one of the data science / machine learning communities. It has a ton of different data sets, challenges that pay big prizes if you win them, challenges to practice with, solution to challenges posted by others in the community, integrated IDEs (based on Google Colab, it’s owned by Google nowadays), and a variety of courses to get started from scratch.

• All Competitions, or first look at the Getting Started competitions to get an idea how it works
• Do anything you feel like with their data sets
• Their courses offer all the necessary skills you need to start your applied data science hobby/career. (Except for underlying probability theory and linalg required to understand the core of the ML algorithms)