Simulation and Validation of Temporal Neural Networks
G. Lin (TU Delft - Electrical Engineering, Mathematics and Computer Science)
G. Gaydadjiev – Mentor (TU Delft - Computer Engineering)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
This thesis explores the simulation and validation of Temporal Neural Networks (TNNs), a form of Artificial Neural Network (ANN) which is relatively underdeveloped, presenting opportunities for further exploration and innovation. TNNs are an interesting research topic, because they attempt to mimic the way biological neural networks process information, relying on temporal pulses or spikes rather than continuous activation. Past works have developed simulators for Temporal Neural Network (TNN) systems, but these simulators often face significant limitations due to the environments in which they are implemented. The primary challenge lies in their inefficiency, which makes conducting large-scale tests difficult. This gives the research question of this thesis is: Can a simulator be developed in an environment that would enable large-scale experimentation and testing of TNN systems? This thesis presents a compiled code event driven simulator for TNN systems, with results for a pre-trained network matching past work, as well as preliminary results for reinforcement learning using TNN systems.