Coping with complexity: machine learning optimization of highly synergistic biological and biochemical systems Mark A. Bedau Professor of Philosophy and Humanities at Reed College   Date: Friday, July 09, 2010 8:30-10:10am

Abstract:
Biological and biochemical systems typically contain complex chemical reactions with many nonlinearities and synergies.
This complexity often prevents a complete understanding of the system, particularly with regard to the prediction of system properties, as experimental parameters such as concentrations, temperature, etc. are changed. This talk describes how a number of different biological and biochemical systems (drug combinations, drug formulations, and protein synthesis) were optimized through a series of iterated high-throughput experiments that are guided by a machine-learning algorithm implementing a form of evolutionary design of experiments. The algorithm predicts fruitful experiments from statistical models of the previous experimental results, combined with stochastic exploration of the experimental space.
These results demonstrate how experimenting on only a tiny but intelligently chosen fraction of the experimental space can very quickly significantly increase the desired response. This evolutionary design of experiments demonstrates the capability for significant innovation, as well as gradual improvement. It is to be expected that continually growing interest in complex experiments, combined with continued improvement in automation of high-throughput experimentation through use of laboratory robotics, will lead to widespread adoption of this approach.

Biosketch:
Prof. Mark A. Bedau (Ph.D. Philosophy, UC Berkeley, 1985; Professor of Philosophy and Humanities at Reed College) is an internationally recognized leader in the philosophical and scientific study of living systems. He has published and lectured around the world extensively on philosophical and scientific issues concerning emergence, evolution, life, mind, and the social and ethical implications of creating life from nonliving materials.

Because he combines training in analytical philosophy with two decades of experience in artificial life, he is a uniquely qualified expert in the philosophical foundations of the life sciences, and has published over 100 research papers, co-authored or co-edited 10 books, including Emergence: Contemporary Readings in Philosophy and Science (MIT Press), The Nature of Life (Cambridge University Press), Protocells: Bridging Nonliving and Living Matter (MIT Press), The Prospect of Protocells: Social and Ethical Implications of Creating Life from Scratch (MIT Press). He has given over 200 lectures in more than 20 countries to audiences in artificial life, computer science, biology, philosophy, cognitive science, psychology, economics, physics, and mathematics, on a variety of philosophical and scientific topics including emergence, evolution, life, mind, and the social and ethical implications of creating life from scratch. He is Editor-in-Chief of the journal Artificial Life (published by MIT Press), and co-organized the last 6 international conference on artificial life. Most recently, he co-founded ProtoLife, Inc., a start-up company with the long-term aim of creating useful artificial cells. He simultaneously co-founded the European Center for Living Technology, a research institute in Venice, Italy, that investigates theoretical and practical issues associated with living systems. He is a regular Visiting Professor at the European School for Molecular Medicine (Milan, Italy), and also at the University of Southern Denmark (Odense, Denmark).

Combinatorial evolution in technology and an algorithm this suggests W. Brian Arthur External Professor, Santa Fe Institute   Date: Saturday, July 10, 2010 8:30 - 10:10 am

Abstract:
Brian Arthur will talk about his new book, The Nature of Technology, which lays out an understanding of how technology comes into being and how it evolves. He will also talk about a new algorithm based on technological evolution that builds up families of technologies from ones that previously exist; and discuss how it compares with genetic algorithms.

Biosketch:
Brian Arthur´s background is in engineering and mathematics, but he is best known as an economist. From 1983 to 1996 he was Dean and Virginia Morrison Professor of Population Studies and Economics at Stanford. And from 1988 to 2004 he was Citibank Professor at the Santa Fe Institute. Arthur is well-known for his “theory of increasing returns”, which explains what happens when products that gain market share find it easier to gain further market share, and how such positive feedbacks lock markets in to the domination of one or two players.

Arthur is also one of the pioneers of the science of complexity - the science of how patterns and structures self-organize. He directed the Santa Fe Institute´s first research program in 1988. He is the recipient of the International Schumpeter Prize in Economics, the inaugural Lagrange Prize in Complexity Science, and two honorary doctorates.