Efficient classification with sparsity augmented collaborative representation

详细信息    查看全文

• 作者：Naveed Akhtar^a ; ^{naveed.akhtar@research.uwa.edu.au}Author Vitae ; Faisal Shafait^bAuthor Vitae ; Ajmal Mian^aAuthor Vitae
• 关键词：Multi-class classification ; Sparse representation ; Collaborative representation
• 刊名：Pattern Recognition
• 出版年：2017
• 出版时间：May 2017
• 年：2017
• 卷：65
• 期：Complete
• 页码：136-145
• 全文大小：994 K
• 卷排序：65

文摘

Sparsity of collaborative representation explicitly contributes to classification. Computational gain is achievable in classification without ignoring sparsity. Combining dense and sparse representations is better than using them alone. Efficient classification scheme is proposed based on combined representation. The scheme out-performs the existing techniques both in accuracy and efficiency.