Learning the Parameters of Determinantal Point Process Kernels

Raja Hafiz Affandi, Emily Fox, Ryan Adams, Ben Taskar
Proceedings of the 31st International Conference on Machine Learning, PMLR 32(2):1224-1232, 2014.

Abstract

Determinantal point processes (DPPs) are well-suited for modeling repulsion and have proven useful in applications where diversity is desired. While DPPs have many appealing properties, learning the parameters of a DPP is difficult, as the likelihood is non-convex and is infeasible to compute in many scenarios. Here we propose Bayesian methods for learning the DPP kernel parameters. These methods are applicable in large-scale discrete and continuous DPP settings, even when the likelihood can only be bounded. We demonstrate the utility of our DPP learning methods in studying the progression of diabetic neuropathy based on the spatial distribution of nerve fibers, and in studying human perception of diversity in images.

Cite this Paper


BibTeX
@InProceedings{pmlr-v32-affandi14, title = {Learning the Parameters of Determinantal Point Process Kernels}, author = {Affandi, Raja Hafiz and Fox, Emily and Adams, Ryan and Taskar, Ben}, booktitle = {Proceedings of the 31st International Conference on Machine Learning}, pages = {1224--1232}, 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/affandi14.pdf}, url = {https://proceedings.mlr.press/v32/affandi14.html}, abstract = {Determinantal point processes (DPPs) are well-suited for modeling repulsion and have proven useful in applications where diversity is desired. While DPPs have many appealing properties, learning the parameters of a DPP is difficult, as the likelihood is non-convex and is infeasible to compute in many scenarios. Here we propose Bayesian methods for learning the DPP kernel parameters. These methods are applicable in large-scale discrete and continuous DPP settings, even when the likelihood can only be bounded. We demonstrate the utility of our DPP learning methods in studying the progression of diabetic neuropathy based on the spatial distribution of nerve fibers, and in studying human perception of diversity in images.} }
Endnote
%0 Conference Paper %T Learning the Parameters of Determinantal Point Process Kernels %A Raja Hafiz Affandi %A Emily Fox %A Ryan Adams %A Ben Taskar %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-affandi14 %I PMLR %P 1224--1232 %U https://proceedings.mlr.press/v32/affandi14.html %V 32 %N 2 %X Determinantal point processes (DPPs) are well-suited for modeling repulsion and have proven useful in applications where diversity is desired. While DPPs have many appealing properties, learning the parameters of a DPP is difficult, as the likelihood is non-convex and is infeasible to compute in many scenarios. Here we propose Bayesian methods for learning the DPP kernel parameters. These methods are applicable in large-scale discrete and continuous DPP settings, even when the likelihood can only be bounded. We demonstrate the utility of our DPP learning methods in studying the progression of diabetic neuropathy based on the spatial distribution of nerve fibers, and in studying human perception of diversity in images.
RIS
TY - CPAPER TI - Learning the Parameters of Determinantal Point Process Kernels AU - Raja Hafiz Affandi AU - Emily Fox AU - Ryan Adams AU - Ben Taskar BT - Proceedings of the 31st International Conference on Machine Learning DA - 2014/06/18 ED - Eric P. Xing ED - Tony Jebara ID - pmlr-v32-affandi14 PB - PMLR DP - Proceedings of Machine Learning Research VL - 32 IS - 2 SP - 1224 EP - 1232 L1 - http://proceedings.mlr.press/v32/affandi14.pdf UR - https://proceedings.mlr.press/v32/affandi14.html AB - Determinantal point processes (DPPs) are well-suited for modeling repulsion and have proven useful in applications where diversity is desired. While DPPs have many appealing properties, learning the parameters of a DPP is difficult, as the likelihood is non-convex and is infeasible to compute in many scenarios. Here we propose Bayesian methods for learning the DPP kernel parameters. These methods are applicable in large-scale discrete and continuous DPP settings, even when the likelihood can only be bounded. We demonstrate the utility of our DPP learning methods in studying the progression of diabetic neuropathy based on the spatial distribution of nerve fibers, and in studying human perception of diversity in images. ER -
APA
Affandi, R.H., Fox, E., Adams, R. & Taskar, B.. (2014). Learning the Parameters of Determinantal Point Process Kernels. Proceedings of the 31st International Conference on Machine Learning, in Proceedings of Machine Learning Research 32(2):1224-1232 Available from https://proceedings.mlr.press/v32/affandi14.html.

Related Material