Pairwise Supervised Hashing with Bernoulli Variational Auto-Encoder and Self-Control Gradient Estimator

Siamak Zamani Dadaneh, Shahin Boluki, Mingzhang Yin, Mingyuan Zhou, Xiaoning Qian
Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI), PMLR 124:540-549, 2020.

Abstract

Semantic hashing has become a crucial component of fast similarity search in many large-scale information retrieval systems, in particular, for text data. Variational auto-encoders (VAEs) with binary latent variables as hashing codes provide state-of-the-art performance in terms of precision for document retrieval. We propose a pairwise loss function with discrete latent VAE to reward within-class similarity and between-class dissimilarity for supervised hashing. Instead of solving the optimization for training relying on existing biased gradient estimators, an unbiased, low-variance gradient estimator, which evaluates the non-differentiable loss function over two correlated sets of binary hashing codes to control the gradient variance, is adopted to optimize the hashing function to achieve superior performance compared to the state-of-the-arts, as demonstrated by our comprehensive experiments.

Cite this Paper

BibTeX
@InProceedings{pmlr-v124-zamani-dadaneh20a, title = {Pairwise Supervised Hashing with Bernoulli Variational Auto-Encoder and Self-Control Gradient Estimator}, author = {Zamani Dadaneh, Siamak and Boluki, Shahin and Yin, Mingzhang and Zhou, Mingyuan and Qian, Xiaoning}, booktitle = {Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI)}, pages = {540--549}, year = {2020}, editor = {Peters, Jonas and Sontag, David}, volume = {124}, series = {Proceedings of Machine Learning Research}, month = {03--06 Aug}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v124/zamani-dadaneh20a/zamani-dadaneh20a.pdf}, url = {https://proceedings.mlr.press/v124/zamani-dadaneh20a.html}, abstract = {Semantic hashing has become a crucial component of fast similarity search in many large-scale information retrieval systems, in particular, for text data. Variational auto-encoders (VAEs) with binary latent variables as hashing codes provide state-of-the-art performance in terms of precision for document retrieval. We propose a pairwise loss function with discrete latent VAE to reward within-class similarity and between-class dissimilarity for supervised hashing. Instead of solving the optimization for training relying on existing biased gradient estimators, an unbiased, low-variance gradient estimator, which evaluates the non-differentiable loss function over two correlated sets of binary hashing codes to control the gradient variance, is adopted to optimize the hashing function to achieve superior performance compared to the state-of-the-arts, as demonstrated by our comprehensive experiments.} }
Endnote
%0 Conference Paper %T Pairwise Supervised Hashing with Bernoulli Variational Auto-Encoder and Self-Control Gradient Estimator %A Siamak Zamani Dadaneh %A Shahin Boluki %A Mingzhang Yin %A Mingyuan Zhou %A Xiaoning Qian %B Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI) %C Proceedings of Machine Learning Research %D 2020 %E Jonas Peters %E David Sontag %F pmlr-v124-zamani-dadaneh20a %I PMLR %P 540--549 %U https://proceedings.mlr.press/v124/zamani-dadaneh20a.html %V 124 %X Semantic hashing has become a crucial component of fast similarity search in many large-scale information retrieval systems, in particular, for text data. Variational auto-encoders (VAEs) with binary latent variables as hashing codes provide state-of-the-art performance in terms of precision for document retrieval. We propose a pairwise loss function with discrete latent VAE to reward within-class similarity and between-class dissimilarity for supervised hashing. Instead of solving the optimization for training relying on existing biased gradient estimators, an unbiased, low-variance gradient estimator, which evaluates the non-differentiable loss function over two correlated sets of binary hashing codes to control the gradient variance, is adopted to optimize the hashing function to achieve superior performance compared to the state-of-the-arts, as demonstrated by our comprehensive experiments.
APA
Zamani Dadaneh, S., Boluki, S., Yin, M., Zhou, M. & Qian, X.. (2020). Pairwise Supervised Hashing with Bernoulli Variational Auto-Encoder and Self-Control Gradient Estimator. Proceedings of the 36th Conference on Uncertainty in Artificial Intelligence (UAI), in Proceedings of Machine Learning Research 124:540-549 Available from https://proceedings.mlr.press/v124/zamani-dadaneh20a.html.

Related Material