IPAB Workshop - 15/10/2020

Yurdusev Yakup Akan

Title: Towards a Theory of Turing Learning 

Description:  Turing learning is a method that aims to model the behaviour observed. In the method, the behaviour of replica agent is evolved towards that of original ones. The fitness of model is proportional to the replica's ability to trick classifier which observes all the individuals in the environment whereas the fitness of classifier depends on distinguishing correctly these two groups of agents. TL allows the machine to infer the behavioural rules autonomously in the homogeneous swarm. I will demonstrate experimental results in which coevolutionary algorithm used and mention about theoretical aspect of learning process.