Latent Semantic Representation Learning for Scene Classification

Xin Li, Yuhong Guo
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):532-540, 2014.

Abstract

The performance of machine learning methods is heavily dependent on the choice of data representation. In real world applications such as scene recognition problems, the widely used low-level input features can fail to explain the high-level semantic label concepts. In this work, we address this problem by proposing a novel patch-based latent variable model to integrate latent contextual representation learning and classification model training in one joint optimization framework. Within this framework, the latent layer of variables bridge the gap between inputs and outputs by providing discriminative explanations for the semantic output labels, while being predictable from the low-level input features. Experiments conducted on standard scene recognition tasks demonstrate the efficacy of the proposed approach, comparing to the state-of-the-art scene recognition methods.

Cite this Paper

BibTeX
@InProceedings{pmlr-v32-lid14, title = {Latent Semantic Representation Learning for Scene Classification}, author = {Li, Xin and Guo, Yuhong}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {532--540}, year = {2014}, editor = {Xing, Eric P. and Jebara, Tony}, volume = {32}, number = {2}, series = {Proceedings of Machine Learning Research}, address = {Bejing, China}, month = {22--24 Jun}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v32/lid14.pdf}, url = {https://proceedings.mlr.press/v32/lid14.html}, abstract = {The performance of machine learning methods is heavily dependent on the choice of data representation. In real world applications such as scene recognition problems, the widely used low-level input features can fail to explain the high-level semantic label concepts. In this work, we address this problem by proposing a novel patch-based latent variable model to integrate latent contextual representation learning and classification model training in one joint optimization framework. Within this framework, the latent layer of variables bridge the gap between inputs and outputs by providing discriminative explanations for the semantic output labels, while being predictable from the low-level input features. Experiments conducted on standard scene recognition tasks demonstrate the efficacy of the proposed approach, comparing to the state-of-the-art scene recognition methods.} }
Endnote
%0 Conference Paper %T Latent Semantic Representation Learning for Scene Classification %A Xin Li %A Yuhong Guo %B Proceedings of the 31st International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2014 %E Eric P. Xing %E Tony Jebara %F pmlr-v32-lid14 %I PMLR %P 532--540 %U https://proceedings.mlr.press/v32/lid14.html %V 32 %N 2 %X The performance of machine learning methods is heavily dependent on the choice of data representation. In real world applications such as scene recognition problems, the widely used low-level input features can fail to explain the high-level semantic label concepts. In this work, we address this problem by proposing a novel patch-based latent variable model to integrate latent contextual representation learning and classification model training in one joint optimization framework. Within this framework, the latent layer of variables bridge the gap between inputs and outputs by providing discriminative explanations for the semantic output labels, while being predictable from the low-level input features. Experiments conducted on standard scene recognition tasks demonstrate the efficacy of the proposed approach, comparing to the state-of-the-art scene recognition methods.
RIS
TY - CPAPER TI - Latent Semantic Representation Learning for Scene Classification AU - Xin Li AU - Yuhong Guo BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-lid14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 532 EP - 540 L1 - http://proceedings.mlr.press/v32/lid14.pdf UR - https://proceedings.mlr.press/v32/lid14.html AB - The performance of machine learning methods is heavily dependent on the choice of data representation. In real world applications such as scene recognition problems, the widely used low-level input features can fail to explain the high-level semantic label concepts. In this work, we address this problem by proposing a novel patch-based latent variable model to integrate latent contextual representation learning and classification model training in one joint optimization framework. Within this framework, the latent layer of variables bridge the gap between inputs and outputs by providing discriminative explanations for the semantic output labels, while being predictable from the low-level input features. Experiments conducted on standard scene recognition tasks demonstrate the efficacy of the proposed approach, comparing to the state-of-the-art scene recognition methods. ER -
APA
Li, X. & Guo, Y.. (2014). Latent Semantic Representation Learning for Scene Classification. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):532-540 Available from https://proceedings.mlr.press/v32/lid14.html.

Related Material