APSIPA Transactions on Signal and Information Processing > Vol 13 > Issue 1

Image Classification via Subspace Learning Machine with Soft Partitioning (SLM/SP)

Hongyu Fu, University of Southern California, USA, hongyufu@usc.edu , Xinyu Wang, University of Southern California, USA, Vinod K. Mishra, DEVCOM Army Research Laboratory, USA, C.-C. Jay Kuo, University of Southern California, USA
 
Suggested Citation
Hongyu Fu, Xinyu Wang, Vinod K. Mishra and C.-C. Jay Kuo (2024), "Image Classification via Subspace Learning Machine with Soft Partitioning (SLM/SP)", APSIPA Transactions on Signal and Information Processing: Vol. 13: No. 1, e13. http://dx.doi.org/10.1561/116.20240003

Publication Date: 21 Aug 2024
© 2024 H. Fu, Z. Wang, V. K. Mishra and C.-C. J. Kuo
 
Subjects
Classification and prediction,  Learning and statistical methods,  Object and scene recognition,  Pattern recognition and learning
 
Keywords
Subspace learning machineimage classificationmachine learningsoft decision tree
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 151 times

In this article:
Introduction 
Background Review 
SLM Tree with Soft Partition (SLM/SP) 
Image Classification with SLM/SP 
Experiments 
Conclusion and Future Work 
Acknowledgments 
References 

Abstract

Subspace partitioning plays a fundamental role in the design of effective classification methods. A novel subspace learning machine (SLM) was recently proposed. It projects feature vectors into a 1D feature subspace and partitions it into two disjoint sets. To effectively generalize the SLM method to high-dimensional feature space, SLM with soft partitioning, denoted by SLM/SP, is proposed in this work. By incorporating the Soft Decision Tree (SDT) data structure for decision learning, the SLM/SP begins with the adaptive learning of a tree structure using local greedy subspace partitioning. Once the tree structure is finalized, all parameters are globally updated. To apply SLM/SP to image classification tasks, we propose modulated designs for the topology of the SDT and a novel module for efficient local representation learning in the subspace learning diagram. The SLM/SP methodology offers efficient training, high classification accuracy, and small model size, underscored by experimental results on image classification benchmarks.

DOI:10.1561/116.20240003