4th year Ph.D. student

Statistics Department
The Wharton School
University of Pennsylvania

E-mail:


Research Interests

Determinantal Point Processes (DPPs) are random point processes well-suited for modelling repulsion. In machine learning and statistics, DPPs are a natural model for subset selection problems where diversity is desired. For example, they can be used to select diverse sets of sentences to form document summaries, or to return relevant but varied text and image search results, or to detect non-overlapping multiple object trajectories in video. Among many remarkable properties, they offer tractable algorithms for exact inference, including computing marginals, computing certain conditional probabilities, and sampling.

In our recent work, we extended these algorithms to approximately infer non-linear DPPs defined over a large amount of data, as well as DPPs defined on continuous spaces using low-rank approximations. We showed that the errors resulting from these approximations can be bounded and that these approximated DPPs are useful in practice. We also presented robust Bayesian algorithms to learn the parameters of the DPP kernels that can be generalized to handle large-scale and continuous DPPs. We also developed a novel temporal process that sequentially select multiple diverse subsets across time.

We demonstarted the advantages of our models on several machine learning and statistical tasks: motion capture video summarization, repulsive mixture modelling, synthesizing diverse human poses, classifying diabetic neuropathy based on nerve fiber samples, understanding human judgment of image diversity and information retrieval with user feedback.

My future directions are to design methods for large-scale inference and learning of other statistical models, especially spatio-temporal processes and apply them to problems in machine learning, statistics and social networks.

Publications

Markov Determinantal Point Processes
R. H. Affandi, A. Kulesza, and E. B. Fox. UAI 2012.
pdf supplement bib
@inproceedings{affandi2012markov,
  title =	 {{Markov Determinantal Point Processes}},
  author =	 {Affandi, R. H. and Kulesza, A. and Fox, E. B.},
  booktitle =	 {Proceedings of the 28th Conference on Uncertainty
	          in Artificial Intelligence},
  year =	 2012,
}	

Nystrom Approximation for Large-Scale Determinantal Processes
R. H. Affandi, A. Kulesza, E. B. Fox and B. Taskar. Oral Presentation, AISTATS 2013.
pdf bib
@INPROCEEDINGS{Affandi:AISTATS2013,
		author={Affandi, R.H. and Kulesza, A. and Fox, E.B. and Taskar, B.},
		title={{Nystrom Approximation for Large-Scale Determinantal Processes}},
		booktitle = "Proc. International Conference on Artificial Intelligence and Statistics",
		year = {2013},
		month = {April},
		}
	
}	

Approximate Inference in Continuous Determinantal Point Processes
R. H. Affandi, E. B. Fox and B. Taskar. Spotlight Presentation, NIPS 2013.
pdf supplement bib
@INPROCEEDINGS{Affandi:NIPS2013,
		author={Affandi, R.H. and Fox, E.B. and Taskar, B.},
		title={{Approximate Inference in Continuous Determinantal Point Processes}},
		booktitle = "{Advances in Neural Information Processing",
		year = {2013},
		month = {December},
		}
	
}	

Learning the Parameters of Determinantal Point Process Kernels
R. H. Affandi, E. B. Fox, R. P. Adams and B. Taskar. To appear in ICML 2014.
pdf bib
@INPROCEEDINGS{Affandi:ICML2014,
				author={Affandi, R.H. and Fox, E.B. and Adams, R.P. and Taskar, B.},
				title={{Learning the Parameters of Determinantal Point Process Kernels}},
				booktitle = "Proc. International Conference on Machine Learning",
				year = {2014},
				month = {June},
				}
}	
         

Curriculum Vitae

CV