1.NEGATIVE DEPENDENCE: Stefanie Jegelka00:00
2.Outline05:10
3.Capturing Diversity: Determinantal Point Processes07:03
4.Determinantal Point Processes08:20
5.Combinatorial Examples11:07
6.DPP samples12:49
7.Repulsion in nature13:35
8.STRONGER NOTIONS OF NEGATIVE DEPENDENCE14:18
9.Example17:53
10.Detour: Positive Dependence18:52
11.Towards a theory of negative dependence21:16
12.Generating Polynomial22:54
13.Strongly Rayleigh Measures27:30
14.Real Stable Polynomials30:01
15.Nice properties of SR32:59
16.Algorithmic Implications34:34
17.Other SR measures35:30
18.IMPLICATIONS: SAMPLING FOR DETERMINANTAL POINT PROCESSES39:48
19.Sampling for general SR measures41:57
20.Sampling empirically43:21
21.How many iterations?44:08
22.Sampling: from fixed-cardinality to general SR47:28
23.Summary50:51
24.[Break]51:46
25.SECOND PART: Suvrit Sra1:04:48
26.THEORY & APPLICATIONS1:05:10
27.Computing Partition functions1:07:20
28.Learning a DPP from data1:17:20
29.MLE for learning a DPP1:19:12
30.Speeding up DPP learning1:22:23
31.Open problems: learning1:23:54
32.APPLICATIONS: Recommender systems1:26:18
33.Nystrom approximation1:29:38
34.Neural network pruning1:34:02
35.Outlier detection1:37:35
36.Optimal design & active learning1:39:49
37."Dual" volume sampling1:45:50
38.Other ML applications1:48:37
39.Related work at this conference1:49:36
40.Perspectives: Recent Results!1:50:06
41.Summary and Outlook1:52:52
42.Thanks1:54:57
43.Q & A1:55:34