Spike-Timing Dependent Plasticity for Evolved Robots  

Oleh:

Paolo, Ezequiel A. Di

Jenis: Article from Journal - e-Journal
Dalam koleksi: Adaptive Behavior vol. 10 no. 3-4 (Jul. 2002), page 243-263.
Topik: evoluyionary robotics; spiking neural networks; spike-timing dependent plasticity; activity-dependent synaptic scaling; neural noise; robustness
Fulltext: 243.pdf (1.34MB)

Isi artikel

Plastic spiking neural networks are shyntesized for phototactic robots using evolutionary techniques. Sypnatic plasticity asymetrically depend on the precise relative timing betwen presypnatic and possypnaticspike at the millisecond range and on longer-term activity-dependent regulatoryscaling. Comparative studies has been carried out for different kinds of plastics neural networks with low and high level of neural noise. In all cases, the evolved controllersare highlyrobust againstinternal sypnatic decay and other pertubations. The importance of precise timing of spikes isdemonstrate by ramdomizing the spike trains. In the low neural noise scenario, weight values undergo rhyithmic changes at the mesoscale due to bursting, but during periods of high activity they are finely regulated at te microscale by shyncronous or entrained firing. spike train randomization results in loss of performance in this case. in contrast, in the high neural noise scenario, robots are robust to loss of information in the timing of the spike trains, demonstrating of conterintuitive results that plasticity, which is dependent on precise spike timing, can work even in its absence, provided in behavioral strategies make use robust longer-term invariants of sensorimotor interaction. A comparison with a rate-basedmodel of sypnaticplasticity showsc that undersimilarly noise condition, asymetrics spike-timing dependent plasticity achievesbetter performance by means of efficient reductionin weight variance over time. Performance also presents negative sensitivityto reduce levels of noise, showing that random firing has a functional values.

Opini Anda

Klik untuk menuliskan opini Anda tentang koleksi ini!

Kembali