Please use this identifier to cite or link to this item:
|Title:||Optimisation of learning-to-learn in spiking neural circuits|
|Keywords:||Neural networks (Computer science)|
|Citation:||D'Amato, K. (2019). Optimisation of learning-to-learn in spiking neural circuits (Master's dissertation).|
|Abstract:||Situated at the intersection of artiﬁcial intelligence and theoretical neuroscience, spiking neural networks (SNNs) have proven valuable for modelling and predicting neural phenomena. SNNs provide rich dynamics that are not replicable by conventional neural networks, exhibiting desirable properties such as self-healing, the ability to exploit noise as a resource, and the ability to solve diﬃcult constraint problems like Sudoku or NP-hard problems such as the Travelling Salesman Problem without training. This work aims to discover initial conﬁgurations that lead to more eﬀective and generalisable learning as part of a broader eﬀort to deduce the computational principles that achieve generalisable learning in the human brain. Building on Pecevski & Maass, the work explores density estimation in simulations of SNN winner-take-all (WTA) circuits and similar constructions by using genetic algorithms (GA) and natural evolution strategies (NES) to search for individual optimal conﬁgurations that minimise Kullback-Leibler (KL) divergence between multiple estimate and target distributions. It demonstrates that in some tasks optimal network conﬁgurations outperform published results even though theoretical zero-delay conditions do not hold, that generalisability can be achieved with a signiﬁcant but non-fatal impact in both conditional and joint estimation tasks, that biological plausibility can be pushed by introducing a novel architecture with realistic modiﬁcations that also achieves competitive performance in conditional estimation tasks, and that there exists a negative relationship between synaptic delay and estimation performance.|
|Appears in Collections:||Dissertations - FacICT - 2019|
Dissertations - FacICTAI - 2019
Items in OAR@UM are protected by copyright, with all rights reserved, unless otherwise indicated.