The Power of Asymmetry in Binary Hashing

Abstract

When approximating binary similarity using the hamming distance between short binary hashes, we show that even if the similarity is symmetric, we can have shorter and more accurate hashes by using two distinct code maps. I.e. by approximating the similarity between $x$ and $x^{\prime }$ as the hamming distance between $f(x)$ and $g(x^{\prime })$, for two distinct binary codes $f,g$, rather than as the hamming distance between $f(x)$ and $f(x^{\prime })$.

Accepted to NIPS 2013, 9 pages, 5 figures

Cited by in corpus (12)

• Simultaneous Feature Learning and Hash Coding with Deep Neural Networks
• A Survey on Metric Learning for Feature Vectors and Structured Data
• Deep Ordinal Hashing with Spatial Attention
• On Symmetric and Asymmetric LSHs for Inner Product Search
• Improved Asymmetric Locality Sensitive Hashing (ALSH) for Maximum Inner Product Search (MIPS)
• Making Online Sketching Hashing Even Faster
• Collaborative Learning for Extremely Low Bit Asymmetric Hashing
• Asymmetric Correlation Quantization Hashing for Cross-modal Retrieval
• Self-supervised asymmetric deep hashing with margin-scalable constraint
• ExchNet: A Unified Hashing Network for Large-Scale Fine-Grained Image Retrieval
• Online Hashing with Efficient Updating of Binary Codes
• Clustering, Hamming Embedding, Generalized LSH and the Max Norm