Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation

Sebastian Tschiatschek, Josip Djolonga, Andreas Krause
Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, PMLR 51:770-779, 2016.

Abstract

Modeling diversity of sets of items is important in many applications such as product recommendation and data summarization. Probabilistic submodular models, a family of models including the determinantal point process, form a natural class of distributions, encouraging effects such as diversity, repulsion and coverage. Current models, however, are limited to small and medium number of items due to the high time complexity for learning and inference. In this paper, we propose FLID, a novel log-submodular diversity model that scales to large numbers of items and can be efficiently learned using noise contrastive estimation. We show that our model achieves state of the art performance in terms of model fit, but can be also learned orders of magnitude faster. We demonstrate the wide applicability of our model using several experiments.

Cite this Paper


BibTeX
@InProceedings{pmlr-v51-tschiatschek16, title = {Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation}, author = {Tschiatschek, Sebastian and Djolonga, Josip and Krause, Andreas}, booktitle = {Proceedings of the 19th International Conference on Artificial Intelligence and Statistics}, pages = {770--779}, year = {2016}, editor = {Gretton, Arthur and Robert, Christian C.}, volume = {51}, series = {Proceedings of Machine Learning Research}, address = {Cadiz, Spain}, month = {09--11 May}, publisher = {PMLR}, pdf = {http://proceedings.mlr.press/v51/tschiatschek16.pdf}, url = {https://proceedings.mlr.press/v51/tschiatschek16.html}, abstract = {Modeling diversity of sets of items is important in many applications such as product recommendation and data summarization. Probabilistic submodular models, a family of models including the determinantal point process, form a natural class of distributions, encouraging effects such as diversity, repulsion and coverage. Current models, however, are limited to small and medium number of items due to the high time complexity for learning and inference. In this paper, we propose FLID, a novel log-submodular diversity model that scales to large numbers of items and can be efficiently learned using noise contrastive estimation. We show that our model achieves state of the art performance in terms of model fit, but can be also learned orders of magnitude faster. We demonstrate the wide applicability of our model using several experiments.} }
Endnote
%0 Conference Paper %T Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation %A Sebastian Tschiatschek %A Josip Djolonga %A Andreas Krause %B Proceedings of the 19th International Conference on Artificial Intelligence and Statistics %C Proceedings of Machine Learning Research %D 2016 %E Arthur Gretton %E Christian C. Robert %F pmlr-v51-tschiatschek16 %I PMLR %P 770--779 %U https://proceedings.mlr.press/v51/tschiatschek16.html %V 51 %X Modeling diversity of sets of items is important in many applications such as product recommendation and data summarization. Probabilistic submodular models, a family of models including the determinantal point process, form a natural class of distributions, encouraging effects such as diversity, repulsion and coverage. Current models, however, are limited to small and medium number of items due to the high time complexity for learning and inference. In this paper, we propose FLID, a novel log-submodular diversity model that scales to large numbers of items and can be efficiently learned using noise contrastive estimation. We show that our model achieves state of the art performance in terms of model fit, but can be also learned orders of magnitude faster. We demonstrate the wide applicability of our model using several experiments.
RIS
TY - CPAPER TI - Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation AU - Sebastian Tschiatschek AU - Josip Djolonga AU - Andreas Krause BT - Proceedings of the 19th International Conference on Artificial Intelligence and Statistics DA - 2016/05/02 ED - Arthur Gretton ED - Christian C. Robert ID - pmlr-v51-tschiatschek16 PB - PMLR DP - Proceedings of Machine Learning Research VL - 51 SP - 770 EP - 779 L1 - http://proceedings.mlr.press/v51/tschiatschek16.pdf UR - https://proceedings.mlr.press/v51/tschiatschek16.html AB - Modeling diversity of sets of items is important in many applications such as product recommendation and data summarization. Probabilistic submodular models, a family of models including the determinantal point process, form a natural class of distributions, encouraging effects such as diversity, repulsion and coverage. Current models, however, are limited to small and medium number of items due to the high time complexity for learning and inference. In this paper, we propose FLID, a novel log-submodular diversity model that scales to large numbers of items and can be efficiently learned using noise contrastive estimation. We show that our model achieves state of the art performance in terms of model fit, but can be also learned orders of magnitude faster. We demonstrate the wide applicability of our model using several experiments. ER -
APA
Tschiatschek, S., Djolonga, J. & Krause, A.. (2016). Learning Probabilistic Submodular Diversity Models Via Noise Contrastive Estimation. Proceedings of the 19th International Conference on Artificial Intelligence and Statistics, in Proceedings of Machine Learning Research 51:770-779 Available from https://proceedings.mlr.press/v51/tschiatschek16.html.

Related Material