1. What are the contributions mentioned in the paper "Supervised binary hash code learning with jensen shannon divergence" ?
This paper proposes to learn binary hash codes within a statistical learning framework, in which an upper bound of the probability of Bayes decision errors is derived for different forms of hash functions and a rigorous proof of the convergence of the upper bound is presented.. This paper also illustrates a fast hash coding method that exploits simple binary tests to achieve orders of magnitude improvement in coding speed as compared to projection based methods.
read more
2. What are the future works in "Supervised binary hash code learning with jensen shannon divergence" ?
As for future work, the authors advocate the JSD-based learning approach as a generic framework to combine different learning algorithms.. In order to alleviate the burden of providing labels for all data points, the authors are interested in a semisupervised approach where the second term of JSD ( in eq. 3 ) can be estimated using only partially labelled data.
read more
![Figure 4. Left to right: Performance comparison of JSD with SH [26], BRE [7], ITQ [3] and KSH [12] on CIFAR10 dataset with Euclidean neighbourhood. Upper: Precision vs number of bits. Bottom: Precision vs Log(recall) with 32 bits code.](/figures/figure-4-left-to-right-performance-comparison-of-jsd-with-sh-jmfddap4.webp)