1. Machine learning-Evolution Computational Model
Nischal S

2. Introduction
Evolutionary Computation is the field of study devoted to the design, development, and analysis is problem solvers based on natural selection (simulated evolution).
Evolutionary programming was introduced by Lawrence J. Fogel in the US, while John Henry Holland called his method a genetic algorithm.
Evolution has proven to be a powerful search process.
A computational model of evolution and suggested that Darwinian evolution be studied in the framework of computational learning theory.
An evolution as a restricted form of learning where exploration is limited to a set of possible mutations and feedback is received through the survival of the organisms that adapt to environment using mutation.

3. Introduction(cntd..)
Darwinian evolution is one of the most important scientific theories and suggests that complex life-forms emerged from simpler ones. Yet, the nature of the complexity that can evolve in organisms and the processes therein are not well understood. The two central aspects of Darwin’s theory are
1) creation of variation due to mutations , and
2)natural selection
among the variants, a.k.a. survival of the organism that adapt best to environment surrounding them.
Underlying DNA sequence or genome of an organism contains code for proteins and also encodes rules governing their regulation.
The genome almost entirely controls the functions of an individual organism. Eg: a function encoded in the genome of an organism could be a circuit that decides the level of enzyme activity based on the environmental conditions (e.g.temperature, presence of oxygen etc.)
Captures the central ideas of mutation(random variation) and natural selection. Understanding evolution in the framework of computational learning theory, and understand the evolutionary process as a restricted form of learning.

4. Cellular level
The goal of computational learning theory is to separate concept classes that can be efficiently learned (ideal) from those that cannot.
Quantify the notion of complexity by mathematical functions, they realize.
Mathematical function - Protein expression level: y=f(x1,x2,x3,x4..xn)
Where x1,x,2,x3 are concentration level of each molecules. What are those functions, how complex are they? How do they succeed in finding optimal expression level given environment factors?
Notion of Ideal Behavior(fi): Optimal Expression level for every possible conditions.
Performance: How close actual fn is to Ideal functions fi? Again this depends on environment conditions distributed over x1,x2,x3..xn. Expresses the fitness function.

5. Functions to evolvable? (constrains)
Reasonable size of population
Reasonable number of generations.
Not consideration of population dynamism(allele frequency)
We find to quantify the process of evolution through ideal function, performance, representation, mutation, selection, goal, evolvability and few other functions.

6. Questions: The compare few models in computation learning theory :
probably approximate correct (PAC) learning framework[1]
Kearns’ statistical query learning (SQ) framework
Correlational statistical query (CSQ) learning framework[2]
Valiant’s model of evolution[3].
The variants introduced by Feldman [4], Fisher [5], Muller [6] and P. Valiant [7] are compared.
We study and understand the nature of complexity that can emerge in these genetic circuits, by understanding existing models from computational learning theory. The connection of evolvability to statistical query learning. We define the merits, accelerations and limitations of each of these models.

7. Applications Routing in Communications Networks Robotics
Machine Learning
Routing in Communications Networks
Robotics
Pattern Recognition
VLSI Circuit Layout
Market Forecasting
Path Planning

8. Bibliography
[1] Leslie G. Valiant. A theory of the learnable. Communications of the ACM, 1984.
[2] Michael J. Kearns. Sq learning. Journal of Computing, 1998.
[3] Leslie G. Valiant. Evolvability. Journal of the ACM, Earlier version appears as Leslie G. Valiant. Evolvability. ECCC Technical Report TR
[4]. Vitaly Feldman. Robustness of evolvability. In Proceedings of the Conference on Learning Theory (COLT), 2009.
[5]. R. A. Fisher. The genetical theory of natural selection. Clarendon Press, 1930.
[6]. H. J. Muller. Some genetic aspects. The American Naturalist, 66(703):118– 138, 1932.
[7]. Paul Valiant. Evolvability of real-valued functions. In Proceedings of Innovations in Theoretical Computer Science (ITCS), 2012.