What makes an optimization problem hard?
Complexity
Kolmogorov Random Graphs and the Incompressibility Method
SIAM Journal on Computing
Information perspective of optimization
PPSN'06 Proceedings of the 9th international conference on Parallel Problem Solving from Nature
No free lunch theorems for optimization
IEEE Transactions on Evolutionary Computation
Hi-index | 0.00 |
Information is the tennis ball of communication- this sentence concluded Keith Devlin's talk trying to answer the question: 'Does information Really Exist?'. Whether information exists or not, the discussion about information surely does. Even though the term information has a formal definition (i.e., Kolmogorov complexity, Shannon's information theory), in the EC community (and, as it seems, in many other fields), the term information is used, more often than not, in an informal way (e.g., the problem contains no information and hence it cannot be solved efficiently). This tutorial focuses on the connection between the two formal definitions of information (Shannnon's entropy and Kolmogorov's Complexity) to Optimisation problems and Black- Box algorithms. For example, we will discuss how informal observations of the kind, 'ONEMAX contains good information', 'NIAH does not contain any' connects with the formal definition. Moreover, we will show that the No Free Lunch Theorems are just a private case of a more general phenomenon. The tutorial covers the following major issues: Kolmogorov complexity (KC) and its relation to Shannon information theory, KC and problem hardness, the relation between KC and other (applicable) predictive measures to problem difficulty (e.g., auto-correlation, ruggedness), an information perspective on the no free- lunch theorems and (if time allows) philosophical aspects of information.