Hash-Based Support Vector Machines Approximation for Large Scale Prediction

Abstract

How-to train effective classifiers on huge amount of multimedia data is clearly a major challenge that is attracting more and more research works across several communities. Less efforts however are spent on the counterpart scalability issue: how to apply big trained models efficiently on huge non annotated media collections ? In this paper, we address the problem of speeding-up the prediction phase of linear Support Vector Machines via Locality Sensitive Hashing. We propose building efficient hashbased classifiers that are applied in a first stage in order to approximate the exact results and filter the hypothesis space. Experiments performed with millions of one-against-one classifiers show that the proposed hash-based classifier can be more than two orders of magnitude faster than the exact classifier with minor losses in quality.

Scalable Mobile Visual Classification by <newline/>Kernel Preserving Projection Over <newline/>High-Dimensional Features

“…Recently, efforts have also been made to improve the testing efficiency. In [16], the authors speed up the classification efficiency given a big trained model which may contain millions of classifiers by hashing both classifiers and features to Hamming space and use the Hamming distance to approximate the original classifier. By replacing inner product on high dimensional floating point vector with compact hamming code, the classification process can be 20 to 200 times faster than using original classifiers, and the size of the classifier can also be significantly reduced.…”