Compact Hyperplane Hashing with Bilinear Functions

TL;DR

This paper introduces a novel bilinear hyperplane hashing method that produces compact, discriminative hash codes, significantly improving search speed and accuracy in large-scale active learning tasks.

Contribution

It proposes a bilinear hash function for hyperplane hashing and a learning framework to optimize these functions directly from data, enhancing performance over existing methods.

Findings

Achieves higher collision probability with shorter hash codes.

Demonstrates superior search accuracy and speed in large-scale experiments.

Outperforms existing hyperplane hashing methods on datasets with up to one million samples.

Abstract

Hyperplane hashing aims at rapidly searching nearest points to a hyperplane, and has shown practical impact in scaling up active learning with SVMs. Unfortunately, the existing randomized methods need long hash codes to achieve reasonable search accuracy and thus suffer from reduced search speed and large memory overhead. To this end, this paper proposes a novel hyperplane hashing technique which yields compact hash codes. The key idea is the bilinear form of the proposed hash functions, which leads to higher collision probability than the existing hyperplane hash functions when using random projections. To further increase the performance, we propose a learning based framework in which the bilinear functions are directly learned from the data. This results in short yet discriminative codes, and also boosts the search performance over the random projection based solutions. Large-scale active learning experiments carried out on two datasets with up to one million samples demonstrate the overall superiority of the proposed approach.