en
0.25
0.5
0.75
1.25
1.5
1.75
2
Near-Optimal MAP Inference for Determinantal Point Processes
Published on Jan 16, 20135792 Views
Determinantal point processes (DPPs) have recently been proposed as computationally efficient probabilistic models of diverse sets for a variety of applications, including document summarization, imag
Related categories
Chapter list
Near-Optimal Map Inference for Determinantal Point Processes00:00
Image Search: “Jaguar”00:08
Task: Subset Selection00:28
Matched Summarization (1)00:42
Matched Summarization (2)00:59
Quality only01:24
Quality + diversity01:44
Formalizing (1)02:00
Formalizing (2)02:24
Formalizing (3)02:41
Area as a Det (1)02:51
Area as a Det (2)03:07
Volume as a Det (1)03:12
Volume as a Det (2)04:01
Volume as a Det (3)04:06
DPP Inference (1)04:09
DPP Inference (2)04:46
DPP Inference (3)04:52
DPP Inference (4)05:04
DPP Inference (5)05:15
Submodularity to the Rescue (1)05:31
Submodularity to the Rescue (2)05:55
Submodularity to the Rescue (3)06:37
Submodularity to the Rescue (4)07:01
Monotonicity07:17
Prior Work (1)07:48
Prior Work (2)08:27
Image Comparison with Constraints (1)08:39
Image Comparison with Constraints (2)08:53
Image Comparison with Constraints (3)09:12
Chekuri et al. 2011 (1)09:44
Chekuri et al. 2011 (2)10:13
Chekuri et al. 2011 (3)10:33
Chekuri et al. 2011 (4)10:37
Chekuri et al. 2011 (5)10:41
Chekuri et al. 2011 (6)10:56
Softmax Extension (1)11:05
Softmax Extension (2)11:21
Softmax Extension (3)11:34
Softmax Extension (4)11:39
Softmax Extension (5)11:44
Softmax Extension (6)12:06
Concave in all-positive/all-negative directions12:34
Not necessarily concave in other directions12:49
Approximation Guarantee12:57
Baseline13:36
Synthetic Experiments13:47
Effectiveness eval13:55
Efficiency eval (1)14:33
Efficiency eval (2)15:03
Matched Summarization (1)15:29
Matched Summarization (2)15:53
Matched Summarization (3)15:59
Matched Summarization (4)16:29
Matched summary17:08
Performance17:34
Summary (1)17:53
Summary (1)18:18