Sparse Distributed Representations for Words with Thresholded Independent Component Analysis
Jaakko J. Väyrynen, Lasse Lindqvist, Timo Honkela · IEEE International Conference on Neural Networks/IEEE ... International Conference on Neural Networks · 2007
We show that independent component analysis (ICA) can be used to find distributed representations for words that can be further processed by thresholding to produce sparse representations. The applicability of the thresholded ICA representation is compared to singular value decomposition (SVD) in a multiple choice vocabulary task with three data sets.