University of Twente Student Theses

Login

Applying Reservoir Computing to Semantic Networks : simulating the use of a boron dopant cell as a reservoir

Koopman, Robbin (2021) Applying Reservoir Computing to Semantic Networks : simulating the use of a boron dopant cell as a reservoir.

[img] PDF
2MB
Abstract:Semantic analysis requires working with a large corpus of information called a semantic network. Modern-day computers are ill-equipped for dealing with this amount of data. Therefore, solutions are being researched in neuromorphic computing. Chen et al. (2020) have developed a solution, namely a boron dopant cell that is capable of solving Boolean logic gates. The goal of the current research is to find out whether such a system could be utilized to run a semantic network. This is done by simulating the cell using reservoir computing and applying it to aspects of the Parallel Distributed Processing (PDP) model (McClelland and Rogers, 2003). In the first four experiments, different structures are tested using one or more reservoirs and control nodes for learning. The results show moderate to high success for linear gates, but a low success rate for non-linear gates due to unequal contribution of reservoirs. In the fifth experiment, saturation cells are added to counter this problem. Lastly, the PDP model is implemented by using one reservoir per attribute, each solving the AND gate. When using the same seed configuration for each reservoir, there is a moderate success rate, but using unique reservoirs does not result in successful iterations.
Item Type:Essay (Master)
Faculty:BMS: Behavioural, Management and Social Sciences
Subject:77 psychology
Programme:Psychology MSc (66604)
Link to this item:https://purl.utwente.nl/essays/88242
Export this item as:BibTeX
EndNote
HTML Citation
Reference Manager

 

Repository Staff Only: item control page