Evolution 1/1/2020

Organizers:

Adi Livnat, Dpt. of Evolutionary and Environmental Biology, University of Haifa

Arnon Lotem, The George S. Wise Faculty of Life Sciences, Tel Aviv University

Christos Papadimitriou, Columbia

Speakers:

Leslie Valiant (Harvard University), Christos Papadimitriou (Columbia University), Lee Altenberg (University of Hawaii), Arnon Lotem (Tel Aviv University), Adi Livnat (University of Haifa), Yoav Ram (IDC Herzliya), Lilach Hadany (Tel Aviv University) and Tzahi Pilpel (The Weizmann Institute).

Location:

Steinhardt Museum (Main Auditorium - GF)

Abstract

Physical, chemical and biological processes that govern natural life forms can be viewed as a model of computation. Examining these processes through that lens may explain phenomena that would be otherwise too complex to grasp. The countless stunning examples of natural life processes that are best described as algorithms range from relatively simple quorum-sensing in bacteria to the complexity of the human brain. Furthermore, the evolutionary process by which these natural algorithms were created is by itself a long-term transformation that may greatly benefit from being studied as a computation.

Yet, some fundamental questions remain: is the probability of mutations occurring an independent random variable? If the rate of mutations in each specific location is not an independent, random variable, one may think of the evolution process as a distributed algorithm, in which one part of the system takes an action according to partial information which causes actions in other parts of the system, which ultimately determines the global state of the system --- the outcome of the computation.

The process may be similar to a learning process in the theoretical computer science sense of Probably-Approximately-Correct or PAC-learning: a small number of accesses to the data are allowed, and the algorithm is able (for certain classes of data) to reconstruct a short global representation of all the data, one which nevertheless deviates only slightly from the true data. Such complex processes which govern evolution, once understood, will allow us to formulate new computation models, which will possibly impact the Theory of Computing, and subsequently also High Tech.

These are only a few of the directions worth exploring in our re-thinking of the Math and Computing aspects of evolution.

Program

9-9:15 Opening remarks

9:15-9:50 Tzahi Pilpel : On optimal mutation rates

9:55-10:30 Lilach Hadany : Plasticity in variation and its evolutionary implications

Coffee break 20 min

10:50-11:35 Christos Papadimitriou: On evolution and machine learning

11:40-12:15 Arnon Lotem : Co-evolution of learning and data acquisition mechanisms: a model for cognitive evolution

Lunch 12:30-2:00

2:00-2:35 Yoav Ram: Necessity is the mother of invention

2:40-3:25 Lee Altenberg: Interpretability and Incomprehensibility in the Products of Evolution

Coffee break 20 min

3:45-4:20 Adi Livnat: The role of sex in evolution

4:25-5:10 Leslie Valiant: Evolution as Learning