Large-Margin Determinantal Point Processes

Boqing Gong, Wei-Lun Chao USC, Kristen Grauman U. of Texas at Austin, Fei Sha USC
Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, PMLR R13:634-643, 2015.

Abstract

Determinantal point processes (DPPs) offer a powerful approach to modeling diversity in many applications where the goal is to select a diverse subset from a ground set of items. We study the problem of learning the parameters (i.e., the kernel matrix) of a DPP from labeled training data. In this paper, we develop a novel parameter estimation technique particularly tailored for DPPs based on the principle of large margin separation. In contrast to the state-of-the-art method of maximum likelihood estimation of the DPP parameters, our large-margin loss function explicitly models errors in selecting the target subsets, and it can be customized to trade off different types of errors (precision vs. recall). Extensive empirical studies validate our contributions, including applications on challenging document and video summarization, where flexibility in balancing different errors while training the summarization models is indispensable.

Cite this Paper

BibTeX
@InProceedings{pmlr-vR13-gong15a, title = {Large-Margin Determinantal Point Processes}, author = {Gong, Boqing and USC, Wei-Lun Chao and Austin, Kristen Grauman U. of Texas at and USC, Fei Sha}, booktitle = {Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence}, pages = {634--643}, year = {2015}, editor = {Meila, Marina and Heskes, Tom}, volume = {R13}, series = {Proceedings of Machine Learning Research}, month = {12--16 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/r13/main/assets/gong15a/gong15a.pdf}, url = {https://proceedings.mlr.press/r13/gong15a.html}, abstract = {Determinantal point processes (DPPs) offer a powerful approach to modeling diversity in many applications where the goal is to select a diverse subset from a ground set of items. We study the problem of learning the parameters (i.e., the kernel matrix) of a DPP from labeled training data. In this paper, we develop a novel parameter estimation technique particularly tailored for DPPs based on the principle of large margin separation. In contrast to the state-of-the-art method of maximum likelihood estimation of the DPP parameters, our large-margin loss function explicitly models errors in selecting the target subsets, and it can be customized to trade off different types of errors (precision vs. recall). Extensive empirical studies validate our contributions, including applications on challenging document and video summarization, where flexibility in balancing different errors while training the summarization models is indispensable.}, note = {Reissued by PMLR on 04 October 2026.} }
Endnote
%0 Conference Paper %T Large-Margin Determinantal Point Processes %A Boqing Gong %A Wei-Lun Chao USC %A Kristen Grauman U. of Texas at Austin %A Fei Sha USC %B Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence %C Proceedings of Machine Learning Research %D 2015 %E Marina Meila %E Tom Heskes %F pmlr-vR13-gong15a %I PMLR %P 634--643 %U https://proceedings.mlr.press/r13/gong15a.html %V R13 %X Determinantal point processes (DPPs) offer a powerful approach to modeling diversity in many applications where the goal is to select a diverse subset from a ground set of items. We study the problem of learning the parameters (i.e., the kernel matrix) of a DPP from labeled training data. In this paper, we develop a novel parameter estimation technique particularly tailored for DPPs based on the principle of large margin separation. In contrast to the state-of-the-art method of maximum likelihood estimation of the DPP parameters, our large-margin loss function explicitly models errors in selecting the target subsets, and it can be customized to trade off different types of errors (precision vs. recall). Extensive empirical studies validate our contributions, including applications on challenging document and video summarization, where flexibility in balancing different errors while training the summarization models is indispensable. %Z Reissued by PMLR on 04 October 2026.
APA
Gong, B., USC, W.C., Austin, K.G.U.o.T.a. & USC, F.S.. (2015). Large-Margin Determinantal Point Processes. Proceedings of the 31st Conference on Uncertainty in Artificial Intelligence, in Proceedings of Machine Learning Research R13:634-643 Available from https://proceedings.mlr.press/r13/gong15a.html. Reissued by PMLR on 04 October 2026.

Related Material