Download presentation
Presentation is loading. Please wait.
Published byBrendan Pierce Modified over 8 years ago
1
Sum-Product Networks Ph.D. Student : Li Weizhuo 2015.1.14
2
2 Outline Motivation Representation Inference Learning
3
3 Motivation Graphical Models
4
4 Motivation Learning Graphical Model
5
5 Outline Motivation Representation Inference Learning
6
6 Representation What does an SPN mean? How to use SPNs to represent other networks? The Context Specific independence(CSI)
7
7 What Does an SPN mean?
8
8 A Univariate Distribution is a SPN
9
9 A Product of SPNs over a Disjoint Variables is an SPN
10
10 A Weighted Sum of SPNs over the Same variables is an SPN
11
11 How to use SPNs to represent other networks? BN SPN MN SPN Mixture Model SPN
12
12 BN → SPN ???
13
13 BN → SPN
14
14 BN → SPN
15
15 BN → SPN ?????
16
16 MN → SPN
17
17 Mixture Model → SPN or
18
18 The Context Specific Independence(CSI)
19
19 An example in Ontology Matching SPN ( Sims Map| Disjoint 1 ) SPN ( Sims Map| Disjoint 0 ) ??????
20
20 An example in Ontology Matching (Cont) SubClassof Map Disjoint Context-specific independence SPN ( Map(Y1,Y2) Similarities(Y1,Y2))|Disjointwith( Y1,Y2) 1 ) X1 Y1 X2 Y2 Z2
21
21 Outline Motivation Representation Inference Learning
22
22 Inference All marginals are computable in time linear in size of SPN. All MAP states are computable in time linear in size of SPN.
23
23 Compute marginals ????? P(X=0)=? 0.5 1 1 0.9 0.5 0 1 1 11 0.74
24
24 Compute MAP ? ?? ? ? 0.5 0.6 0.4 0.1 0.04 0.3 1 0 1 11 0.12 Max
25
25 Outline Motivation Representation Inference Learning
26
26 Learning Generative weight learning Discriminative weight learning Structure Learning
27
27 Generative weight learning (Poon,H & Domingos, UAI (2011))
28
28 Random forest Hard EM Generative weight learning (Poon,H & Domingos, UAI (2011))
29
29 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
30
30 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
31
31 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
32
32 Discriminative weight learning (Gens,R & Domingos, NIPS(2012)) Bottom-Up
33
33 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
34
34 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
35
35 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
36
36 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
37
37 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
38
38 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
39
39 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
40
40 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
41
41 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
42
42 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
43
43 Discriminative weight learning (Gens,R & Domingos, NIPS(2012))
44
44 Structure Learning (Gens,R & Domingos, ICML(2013)) Mutual information Hard EM
45
45 Summary Maybe Nothing!
46
46 Summary
47
47 References Most of the materials come from Domingo's slides. Source code http://spn.cs.washington.edu/code.shtml video http://videolectures.net/nips2012_gens_discriminative _learning/ http://research.microsoft.com/apps/video/default.aspx ?id=192562&r=1
48
48 Thanks! Q&A
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.