Princeton University Library Catalog

Learning With Memory in Neural Networks

Author/​Artist:
Ho, Katy [Browse]
Format:
Senior thesis
Language:
English
Advisor(s):
Ramadge, Peter J. [Browse]
Department:
Princeton University. Department of Electrical Engineering [Browse]
Certificate:
Princeton University. Program in Applications of Computing [Browse]
Class year:
2017
Summary note:
This paper explores how different parameters of the memory of a recurrent neural network affect learning performance. The size of memory and the effects of long term dependencies are studied in depth for simple, yet interesting problems in which memory plays a central role. Different types of recurrent neural network architectures and structured recurrent neural networks are also explored.