Sparse Distributed Representations for Words with Thresholded Independent Component Analysis
- Resource Type
- Conference
- Authors
- Vayrynen, Jaakko J.; Lindqvist, Lasse; Honkela, Timo
- Source
- 2007 International Joint Conference on Neural Networks Neural Networks, 2007. IJCNN 2007. International Joint Conference on. :1031-1036 Aug, 2007
- Subject
- Computing and Processing
Components, Circuits, Devices and Systems
Signal Processing and Analysis
Independent component analysis
Vocabulary
Testing
Humans
Natural languages
Singular value decomposition
Unsupervised learning
Neural networks
Image coding
Noise reduction
- Language
- ISSN
- 2161-4393
2161-4407
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.