- About the Tutorial:
Machine Learning is a fast growing, rapidly advancing field that touches nearly everyone's lives. There has recently been an explosion of successful machine learning applications - in everything from voice recognition to text analysis to deeper insights for researchers. While common and frequently talked about, most people have only a vague concept of how machine learning actually works.
In this tutorial, Dr. Artemy Kolchinsky and Dr. Brendan Tracey outline exactly what it is that makes machine learning so special in an accessible way. The principles of training and generalization in machine learning are explained with ample metaphors and visual intuitions, an extended analysis of machine learning in games provides a thorough example, and a closer look at the deep neural nets that are the core of successful machine learning. Finally, it addresses when it's appropriate to use (and not use) machine learning in problem solving, as well as an example of scientific research incorporating machine learning principles.
Students of all levels should be able to follow this reasonably-paced introduction to one of the most important engineering breakthroughs of our time.
Read the instructors' interview about this tutorial in our news feed here.
- About the Instructor(s):
Artemy Kolchinsky is a postdoctoral researcher at the Santa Fe Institute. He studies fundamental physical constraints on how information is processed in complex systems, whether in living cells, digital computers, or other processes. He is also using statistical physics to define a notion of semantic information, i.e. information that doesn't simply reflect correlations but rather the amount of meaning for a given system. Artemy holds a PhD from Indiana University, Bloomington with a specialization in complex systems and a minor in cognitive science.
Brendan Tracey works in the overlaps between statistics, machine learning, and engineering analysis in design. He is particularly interested in incorporating non-traditional ideas to more effectively use our simulation tools. Especially in high performance systems, it is difficult to measure the quality of the design, requiring a resource-intensive computer simulation or expensive physical experiment. His research addresses this problem by using data-driven techniques to inform and improve physical models, and by harnessing statistical techniques to learn when low-fidelity models are sufficient and when more accurate high-fidelity models are necessary for engineering design. Outside of these main interests, Brendan is also interested in game theory, graph theory and social choice design.
Brendan received his B.S. in Mechanical Engineering from the University of Rochester and received his M.S. and Ph. D. from Stanford University in the department of Aeronautics and Astronautics. He is currently a postdoc at the Santa Fe Institute and the Department of Aeronautics and Astronautics at the Massachusetts Institute of Technology.
- Types of Machine Learning
- A Geometric View of Supervised Learning
- Generalization Performance
- Artificial Intelligence and Board Games
- Go as a Supervised Learning Problem
- Fundamentals of Game-Playing Systems
- Building a Go Machine Learning Program
- Introduction to Neural Networks
- Why do Deep Neural Networks Succeed?
- The Mystery of Deep Learning
- Some Caveats of Using Machine Learning
- Advanced Concepts: Stacked Monte Carlo