Introduction Motivation and Contribution Motivation Problem: All existing methods use the same number of bits for different projected dimensions with different variances. PCA2 PCAI XX X **+ Possible Solutions: oDifferent number of bits for different dimensions (Unfortunately,have not found an effective way) oIsotropic (equal)variances for all dimensions Li (http://www.cs.sjtu.edu.cn/-livujun) Learning to Hash CSE,SJTU 16 /45
Introduction Motivation and Contribution Motivation Problem: All existing methods use the same number of bits for different projected dimensions with different variances. Possible Solutions: Different number of bits for different dimensions (Unfortunately, have not found an effective way) Isotropic (equal) variances for all dimensions Li (http://www.cs.sjtu.edu.cn/~liwujun) Learning to Hash CSE, SJTU 16 / 45
Introduction Motivation and Contribution Contribution Isotropic hashing (IsoHash):(Kong and Li,2012b) hashing with isotropic variances for all dimensions Multiple-bit quantization: (1)Double-bit quantization(DBQ):(Kong and Li,2012a) Hamming distance driven (2)Manhattan hashing(MH):(Kong et al.,2012) Manhattan distance driven 日卡三4元，互Q0 Li (http://www.cs.sjtu.edu.cn/-livujun Learning to Hash CSE,SJTU 17 /45
Introduction Motivation and Contribution Contribution Isotropic hashing (IsoHash):(Kong and Li, 2012b) hashing with isotropic variances for all dimensions Multiple-bit quantization: (1) Double-bit quantization (DBQ):(Kong and Li, 2012a) Hamming distance driven (2) Manhattan hashing (MH):(Kong et al., 2012) Manhattan distance driven Li (http://www.cs.sjtu.edu.cn/~liwujun) Learning to Hash CSE, SJTU 17 / 45
Isotropic Hashing Outline Introduction oProblem Definition Existing Methods o Motivation and Contribution ②Isotropic Hashing Model Learning o Experimental Results Multiple-Bit Quantization Double-Bit Quantization Manhattan Quantization Conclusion Reference 日卡回24元，互Q0 Li (http://www.cs.sjtu.edu.cn/-livujun) Learning to Hash CSE,SJTU 18 /45
Isotropic Hashing Outline 1 Introduction Problem Definition Existing Methods Motivation and Contribution 2 Isotropic Hashing Model Learning Experimental Results 3 Multiple-Bit Quantization Double-Bit Quantization Manhattan Quantization 4 Conclusion 5 Reference Li (http://www.cs.sjtu.edu.cn/~liwujun) Learning to Hash CSE, SJTU 18 / 45