Evolving plastic neural networks for online learning: review and future directions

摘要: Recent years have seen a resurgence of interest in evolving plastic neural networks for online learning. These approaches an intrinsic appeal --- since, to date, the only working example general intelligence is human brain, which has developed through evolution, and exhibits great capacity adapt unfamiliar environments. In this paper we review prior work area including problem domains tasks, fitness functions, synaptic plasticity models network encoding schemes. We conclude with discussion current findings promising future directions, incorporation functional properties observed biological appear play role learning processes, addressing "general" by introduction previously unseen tasks during evolution process.