Information-Theoretic Dictionary Learning for Image Classification
Authors:
- 11 April 2014
- journal article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 36 (11) , 2173-2184
Abstract
We present a two-stage approach for learning dictionaries for object classification tasks based on the principle of information maximization. The proposed method seeks a dictionary that is compact, discriminative, and generative. In the first stage, dictionary atoms are selected from an initial dictionary by maximizing the mutual information measure on dictionary compactness, discrimination and reconstruction. In the second stage, the selected dictionary atoms are updated for improved reconstructive and discriminative power using a simple gradient ascent algorithm on mutual information. Experiments using real data sets demonstrate the effectiveness of our approach for image classification tasks.Keywords
All Related Versions
Funding Information
- MURI (N00014-10-1-0934)
This publication has 22 references indexed in Scilit:
- Fisher Discrimination Dictionary Learning for sparse representation2011
- Task-Driven Dictionary LearningIEEE Transactions on Pattern Analysis and Machine Intelligence, 2011
- Discriminative K-SVD for dictionary learning in face recognition2010
- Locality-constrained Linear Coding for image classification2010
- Semantic Coding by Supervised Dimensionality ReductionIEEE Transactions on Multimedia, 2008
- Discriminative learned dictionaries for local image analysis2008
- Robust Face Recognition via Sparse RepresentationIEEE Transactions on Pattern Analysis and Machine Intelligence, 2008
- $rm K$-SVD: An Algorithm for Designing Overcomplete Dictionaries for Sparse RepresentationIEEE Transactions on Signal Processing, 2006
- Beyond Bags of Features: Spatial Pyramid Matching for Recognizing Natural Scene Categories2006
- Separability-based multiscale basis selection and feature extraction for signal and image classificationIEEE Transactions on Image Processing, 1998