University of Twente Student Theses
Biologically Realistic Artificial Neural Networks
Kok, M.R. (2020) Biologically Realistic Artificial Neural Networks.
|
PDF
1MB |
| Abstract: | Artificial Neural Networks (ANNs) play an important role in machine learning nowadays. A typical ANN is traditionally trained using stochastic gradient descent (SGD) with backpropagation (BP). However, it is unlikely that a real biological neural network is trained similarly. Neuroscience theories, such as Hebbian Theory could inspire adaption from the traditional training method SGD to make ANNs more biologically plausible. Two mathematical descriptions of Hebbian theory will be suggested based on the limitations of the mathematical framework of Hebbian theory: A competitive Hebbian learning rule and an imply Hebbian learning rule. Finally, this research will propose a method, called the Hebbian Plasticity Term (HPT) method, that incorporates the mathematical description of Hebbian theory to modify the traditional training method SGD. Therefore two variants of the HPT method are proposed: HPT-Competitive and HPT-Imply. The influence of the hebbian plasticity term ϕ on SGD shows more biologically realistic plasticity of a synaptic connection between neurons in the ANN at the cost of performance. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 30 exact sciences in general, 31 mathematics, 50 technical science in general, 54 computer science |
| Programme: | Business & IT BSc (56066) |
| Awards: | Best paper award, Best presentation |
| Link to this item: | https://purl.utwente.nl/essays/82028 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page