SlideShare a Scribd company logo
4.4 Coping with loops
•            DAG

•
                   oscillation:
•

•
             coherent
• coherent
                                      A


                                  B       C
                                              E
                                      D
1. clustering
  –

2. conditioning
  –

3. stochastic simulation
  –
4.4.1 Clustering Methods
A
• B                 C
          Z                                    B              C

    z ∈ {(+b,+c), (¬b,+c), (+b,¬c), (¬b,¬c)}                      E
                                                       D
•

• Z B C                                                A
                                                             P(b,c|a)
                                                   Z
                                                       B,C              P(e|c)
                                                                   E
                                                       D
                                                             P(d|b,c)
•

    –                             (=      )

    –
•

•
    join trees(=clique trees or junction trees)
Join Tree
1.                          NB                      G
     –
     –
     –   →            G NB I-map       I(G)⊆I(NB)
2.   G chordal supergraph G’
     ※3.2.4 graph-traiangulation algorithm
3.   G’ clique                            join tree T
4.   NB
     –                                 T clique-node
     –        clique-node        λ
5.   T                                        clique-node
         NB
G       Join Tree

1. fill-in                                  G’
   1. maximum cardinality search
      •          1   |V|(       )
      •
      •
   2. for (n=|V| to 1) n

2. G’
3.
   C1, C2, ..., Ct
4. Ci Cj(i>j)
Join Tree                                                           Z1

               A                                                                 P(d|b,c)            A,B,C
                                                   A                                                                P(e|c)
                                                                                   Z2
                                                                                                                                   Z3
  B                        C                  B              C                             B,C,D                        C,E

                                   E              D                               λD(z2)                                      λE(z3)
               D                                                         E                 D=0                      E=1

M zi |z j = P(zi | z j ) = P(zi | zi ∩ z j ).     M z2 |z1 = P(z2 | z1 ) = P(b, c, d | b, c) = P(d | b, c)
                                                  M z3|z1 = P(z3 | z1 ) = P(c, e | c) = P(e | c)
           
              0 if      z2 = (b, c, +d)          π (z1 ) = π z (z1 ) = π z (z1 ) = P(z1 ) = P(a, b, c) = P(b | a)P(c | a)P(a)
λD (z2 ) =                                                       2          3


           
              1 if      z2 = (b, c,¬d)           π (z2 ) = ∑ P(z2 | z1 )P(z1 ) = ∑ P(d | b, c)P(a, b, c) = P(d | b, c)P(b, c)
           
              0 if      z3 = (c,¬e)
                                                             Z1                       a

λE (z3 ) =                                       π (z3 ) = ∑ P(z3 | z1 )P(z1 ) = ∑ P(e | c)P(a, b, c) = P(e | c)P(c)
           
              1 if      z3 = (c,+e)                         Z1                      a,b


λZ (z1 ) = M z |z • λD (z2 ) = P(¬d | b, c)
   2               2 1
                                                  BEL(z1 ) = αλZ2 (z1 )λZ3 (z1 )π (z1 )
λZ (z1 ) = M z |z • λE (z3 ) = P(+e | c)
   3               3 1
                                                                      = α P(¬d | b, c)P(+e | c)P(b | a)P(c | a)P(a)
                                                         8                                       BEL(a),BEL(b),BEL(c)
4.4.2 The Method of Conditioning(Reasoning by Assumptions)
Conditioning
                            A                          •
                                                               –       a=0 a=1
                                                       •
                  B                 C
                                                       •           a
                                             E         •                         BEL(b),BEL(c)
                            D                              a

A=0          A=0                A=1              A=1       BEL(b) = P (b | e) = ∑ P(b | a, e) P (a | e)
                                                                                   a
 A            A                 A                 A

 B            C                 B                C

D=0     D              E        D=0      D              E
                      E=1                              E=1
                      w0            w1
w = (w1, w0 ) = P(a)
w E = P(a | +e) = α P(+e | a)P(a)
w E,D = P(a | +e,¬d) = α P(¬d | a, +e)P(a | +e)
A Short Course on Graphical Models
3 The Junction Tree Algoritms – Mark Paskin

http://ai.stanaford.edu/~paskin/gm-short-course/lec3.pdf
Join Tree? ,Junction Tree?
•
  en.wikipedia.org/wiki/Tree_decomposition
• Join Tree                     DAG
• Junction Tree Algorithm
Junction Tree
    G               d                      T                    {b,d}

             b
                            f                                    {b}
        a
                        e




                                                                          {b,e}
                                                        {b,c}
             c                              {a,b,c}             {b,c,e}           {b,e,f}




                  T G Junction Tree
•       singly connected(    ):

•       covering:
        G           A                 C               A⊆C
•       running intersection:
                B C             i         B C                                           i

※Junction Tree Algorithm
Junction Tree
※                   BEL
    λ π
•

•                         (
             )
•
message passing
• message passing protocol:                  B                   C
          C

•                                                                                  (

    1.    Collect(C): for (B <- C.children) { Collect(B) }
    2.    Distribute(C): for (B <- C.children) {Distribute(B)}
• The Shafer-Shenoy Algorithm:
  B    C
    –                           C
    – C
•                                                                {b,d}
    – Lauritzen-Spiegelhalter
    – Hugin                                                      1{b} 4

                                                          2                1




                                                                           {b,e}
                                                        {b,c}

                                              {a,b,c}            {b,c,e}               {b,e,f}
                                                         3                 4

More Related Content

What's hot

A new practical algorithm for volume estimation using annealing of convex bodies
A new practical algorithm for volume estimation using annealing of convex bodiesA new practical algorithm for volume estimation using annealing of convex bodies
A new practical algorithm for volume estimation using annealing of convex bodies
Vissarion Fisikopoulos
 
Clustering in Hilbert simplex geometry
Clustering in Hilbert simplex geometryClustering in Hilbert simplex geometry
Clustering in Hilbert simplex geometry
Frank Nielsen
 
Igv2008
Igv2008Igv2008
Igv2008
shimpeister
 
Divergence center-based clustering and their applications
Divergence center-based clustering and their applicationsDivergence center-based clustering and their applications
Divergence center-based clustering and their applications
Frank Nielsen
 
Engr 371 final exam april 2010
Engr 371 final exam april 2010Engr 371 final exam april 2010
Engr 371 final exam april 2010
amnesiann
 
Computational Information Geometry: A quick review (ICMS)
Computational Information Geometry: A quick review (ICMS)Computational Information Geometry: A quick review (ICMS)
Computational Information Geometry: A quick review (ICMS)
Frank Nielsen
 
Classification with mixtures of curved Mahalanobis metrics
Classification with mixtures of curved Mahalanobis metricsClassification with mixtures of curved Mahalanobis metrics
Classification with mixtures of curved Mahalanobis metrics
Frank Nielsen
 
Montpellier Math Colloquium
Montpellier Math ColloquiumMontpellier Math Colloquium
Montpellier Math Colloquium
Christian Robert
 
Divergence clustering
Divergence clusteringDivergence clustering
Divergence clustering
Frank Nielsen
 
20110602labseminar pub
20110602labseminar pub20110602labseminar pub
20110602labseminar pub
sesejun
 
Engr 371 final exam april 2006
Engr 371 final exam april 2006Engr 371 final exam april 2006
Engr 371 final exam april 2006
amnesiann
 
Comparing estimation algorithms for block clustering models
Comparing estimation algorithms for block clustering modelsComparing estimation algorithms for block clustering models
Comparing estimation algorithms for block clustering models
BigMC
 
Mesh Processing Course : Differential Calculus
Mesh Processing Course : Differential CalculusMesh Processing Course : Differential Calculus
Mesh Processing Course : Differential Calculus
Gabriel Peyré
 
Form 5 formulae and note
Form 5 formulae and noteForm 5 formulae and note
Form 5 formulae and note
smktsj2
 
Patch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective DivergencesPatch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective Divergences
Frank Nielsen
 
Ecfft zk studyclub 9.9
Ecfft zk studyclub 9.9Ecfft zk studyclub 9.9
Ecfft zk studyclub 9.9
Alex Pruden
 
On maximal and variational Fourier restriction
On maximal and variational Fourier restrictionOn maximal and variational Fourier restriction
On maximal and variational Fourier restriction
VjekoslavKovac1
 
Module 5 Sets
Module 5 SetsModule 5 Sets
Module 5 Sets
guestcc333c
 
Mesh Processing Course : Geodesic Sampling
Mesh Processing Course : Geodesic SamplingMesh Processing Course : Geodesic Sampling
Mesh Processing Course : Geodesic Sampling
Gabriel Peyré
 
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
Yong Heui Cho
 

What's hot (20)

A new practical algorithm for volume estimation using annealing of convex bodies
A new practical algorithm for volume estimation using annealing of convex bodiesA new practical algorithm for volume estimation using annealing of convex bodies
A new practical algorithm for volume estimation using annealing of convex bodies
 
Clustering in Hilbert simplex geometry
Clustering in Hilbert simplex geometryClustering in Hilbert simplex geometry
Clustering in Hilbert simplex geometry
 
Igv2008
Igv2008Igv2008
Igv2008
 
Divergence center-based clustering and their applications
Divergence center-based clustering and their applicationsDivergence center-based clustering and their applications
Divergence center-based clustering and their applications
 
Engr 371 final exam april 2010
Engr 371 final exam april 2010Engr 371 final exam april 2010
Engr 371 final exam april 2010
 
Computational Information Geometry: A quick review (ICMS)
Computational Information Geometry: A quick review (ICMS)Computational Information Geometry: A quick review (ICMS)
Computational Information Geometry: A quick review (ICMS)
 
Classification with mixtures of curved Mahalanobis metrics
Classification with mixtures of curved Mahalanobis metricsClassification with mixtures of curved Mahalanobis metrics
Classification with mixtures of curved Mahalanobis metrics
 
Montpellier Math Colloquium
Montpellier Math ColloquiumMontpellier Math Colloquium
Montpellier Math Colloquium
 
Divergence clustering
Divergence clusteringDivergence clustering
Divergence clustering
 
20110602labseminar pub
20110602labseminar pub20110602labseminar pub
20110602labseminar pub
 
Engr 371 final exam april 2006
Engr 371 final exam april 2006Engr 371 final exam april 2006
Engr 371 final exam april 2006
 
Comparing estimation algorithms for block clustering models
Comparing estimation algorithms for block clustering modelsComparing estimation algorithms for block clustering models
Comparing estimation algorithms for block clustering models
 
Mesh Processing Course : Differential Calculus
Mesh Processing Course : Differential CalculusMesh Processing Course : Differential Calculus
Mesh Processing Course : Differential Calculus
 
Form 5 formulae and note
Form 5 formulae and noteForm 5 formulae and note
Form 5 formulae and note
 
Patch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective DivergencesPatch Matching with Polynomial Exponential Families and Projective Divergences
Patch Matching with Polynomial Exponential Families and Projective Divergences
 
Ecfft zk studyclub 9.9
Ecfft zk studyclub 9.9Ecfft zk studyclub 9.9
Ecfft zk studyclub 9.9
 
On maximal and variational Fourier restriction
On maximal and variational Fourier restrictionOn maximal and variational Fourier restriction
On maximal and variational Fourier restriction
 
Module 5 Sets
Module 5 SetsModule 5 Sets
Module 5 Sets
 
Mesh Processing Course : Geodesic Sampling
Mesh Processing Course : Geodesic SamplingMesh Processing Course : Geodesic Sampling
Mesh Processing Course : Geodesic Sampling
 
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
Analysis of an E-plane waveguide T-junction with a quarter-wave transformer u...
 

Viewers also liked

確率伝播
確率伝播確率伝播
確率伝播
Yoshihide Nishio
 
異常検知と変化検知 9章 部分空間法による変化点検知
異常検知と変化検知 9章 部分空間法による変化点検知異常検知と変化検知 9章 部分空間法による変化点検知
異常検知と変化検知 9章 部分空間法による変化点検知
hagino 3000
 
異常行動検出入門(改)
異常行動検出入門(改)異常行動検出入門(改)
異常行動検出入門(改)Yohei Sato
 
条件付き確率場の推論と学習
条件付き確率場の推論と学習条件付き確率場の推論と学習
条件付き確率場の推論と学習Masaki Saito
 
さらば!データサイエンティスト
さらば!データサイエンティストさらば!データサイエンティスト
さらば!データサイエンティスト
Shohei Hido
 
Jubatus Casual Talks #2 異常検知入門
Jubatus Casual Talks #2 異常検知入門Jubatus Casual Talks #2 異常検知入門
Jubatus Casual Talks #2 異常検知入門
Shohei Hido
 
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
Shohei Hido
 
FIT2012招待講演「異常検知技術のビジネス応用最前線」
FIT2012招待講演「異常検知技術のビジネス応用最前線」FIT2012招待講演「異常検知技術のビジネス応用最前線」
FIT2012招待講演「異常検知技術のビジネス応用最前線」
Shohei Hido
 
時系列分析による異常検知入門
時系列分析による異常検知入門時系列分析による異常検知入門
時系列分析による異常検知入門Yohei Sato
 
【プレゼン】見やすいプレゼン資料の作り方【初心者用】
【プレゼン】見やすいプレゼン資料の作り方【初心者用】【プレゼン】見やすいプレゼン資料の作り方【初心者用】
【プレゼン】見やすいプレゼン資料の作り方【初心者用】
MOCKS | Yuta Morishige
 
あなたの業務に機械学習を活用する5つのポイント
あなたの業務に機械学習を活用する5つのポイントあなたの業務に機械学習を活用する5つのポイント
あなたの業務に機械学習を活用する5つのポイント
Shohei Hido
 

Viewers also liked (11)

確率伝播
確率伝播確率伝播
確率伝播
 
異常検知と変化検知 9章 部分空間法による変化点検知
異常検知と変化検知 9章 部分空間法による変化点検知異常検知と変化検知 9章 部分空間法による変化点検知
異常検知と変化検知 9章 部分空間法による変化点検知
 
異常行動検出入門(改)
異常行動検出入門(改)異常行動検出入門(改)
異常行動検出入門(改)
 
条件付き確率場の推論と学習
条件付き確率場の推論と学習条件付き確率場の推論と学習
条件付き確率場の推論と学習
 
さらば!データサイエンティスト
さらば!データサイエンティストさらば!データサイエンティスト
さらば!データサイエンティスト
 
Jubatus Casual Talks #2 異常検知入門
Jubatus Casual Talks #2 異常検知入門Jubatus Casual Talks #2 異常検知入門
Jubatus Casual Talks #2 異常検知入門
 
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
機械学習モデルフォーマットの話:さようならPMML、こんにちはPFA
 
FIT2012招待講演「異常検知技術のビジネス応用最前線」
FIT2012招待講演「異常検知技術のビジネス応用最前線」FIT2012招待講演「異常検知技術のビジネス応用最前線」
FIT2012招待講演「異常検知技術のビジネス応用最前線」
 
時系列分析による異常検知入門
時系列分析による異常検知入門時系列分析による異常検知入門
時系列分析による異常検知入門
 
【プレゼン】見やすいプレゼン資料の作り方【初心者用】
【プレゼン】見やすいプレゼン資料の作り方【初心者用】【プレゼン】見やすいプレゼン資料の作り方【初心者用】
【プレゼン】見やすいプレゼン資料の作り方【初心者用】
 
あなたの業務に機械学習を活用する5つのポイント
あなたの業務に機械学習を活用する5つのポイントあなたの業務に機械学習を活用する5つのポイント
あなたの業務に機械学習を活用する5つのポイント
 

Similar to 確率伝播その2

Past year iit entrance mathematics problems
Past year iit entrance mathematics problemsPast year iit entrance mathematics problems
Past year iit entrance mathematics problems
APEX INSTITUTE
 
Sol7
Sol7Sol7
Sslc maths-5-model-question-papers-english-medium
Sslc maths-5-model-question-papers-english-mediumSslc maths-5-model-question-papers-english-medium
Sslc maths-5-model-question-papers-english-medium
mohanavaradhan777
 
calculo vectorial
calculo vectorialcalculo vectorial
calculo vectorial
Chalio Solano
 
Ch10 29
Ch10 29Ch10 29
Ch10 29
schibu20
 
brain gate
brain gatebrain gate
brain gate
Prince Jairaj
 
Maths`
Maths`Maths`
Maths`
singarls19
 
10th Maths
10th Maths10th Maths
10th Maths
singarls19
 
10th Maths model3 question paper
10th Maths model3 question paper10th Maths model3 question paper
10th Maths model3 question paper
singarls19
 
Bt0063 mathematics fot it
Bt0063 mathematics fot itBt0063 mathematics fot it
Bt0063 mathematics fot it
nimbalkarks
 
Number theory lecture (part 1)
Number theory lecture (part 1)Number theory lecture (part 1)
Number theory lecture (part 1)
Aleksandr Yampolskiy
 
79ecb3d9 65f4-4161-b97d-63711df5d6c5
79ecb3d9 65f4-4161-b97d-63711df5d6c579ecb3d9 65f4-4161-b97d-63711df5d6c5
79ecb3d9 65f4-4161-b97d-63711df5d6c5
spoider
 
Datamining 8th Hclustering
Datamining 8th HclusteringDatamining 8th Hclustering
Datamining 8th Hclustering
sesejun
 
Datamining 8th hclustering
Datamining 8th hclusteringDatamining 8th hclustering
Datamining 8th hclustering
sesejun
 
M A T H E M A T I C S I I I J N T U M O D E L P A P E R{Www
M A T H E M A T I C S  I I I  J N T U  M O D E L  P A P E R{WwwM A T H E M A T I C S  I I I  J N T U  M O D E L  P A P E R{Www
M A T H E M A T I C S I I I J N T U M O D E L P A P E R{Www
guest3f9c6b
 
S101-52國立新化高中(代理)
S101-52國立新化高中(代理)S101-52國立新化高中(代理)
S101-52國立新化高中(代理)
yustar1026
 
Maieee03
Maieee03Maieee03
Maieee03
Ashish Yadav
 
Aieee 2003 maths solved paper by fiitjee
Aieee 2003 maths solved paper by fiitjeeAieee 2003 maths solved paper by fiitjee
Aieee 2003 maths solved paper by fiitjee
Mr_KevinShah
 

Similar to 確率伝播その2 (18)

Past year iit entrance mathematics problems
Past year iit entrance mathematics problemsPast year iit entrance mathematics problems
Past year iit entrance mathematics problems
 
Sol7
Sol7Sol7
Sol7
 
Sslc maths-5-model-question-papers-english-medium
Sslc maths-5-model-question-papers-english-mediumSslc maths-5-model-question-papers-english-medium
Sslc maths-5-model-question-papers-english-medium
 
calculo vectorial
calculo vectorialcalculo vectorial
calculo vectorial
 
Ch10 29
Ch10 29Ch10 29
Ch10 29
 
brain gate
brain gatebrain gate
brain gate
 
Maths`
Maths`Maths`
Maths`
 
10th Maths
10th Maths10th Maths
10th Maths
 
10th Maths model3 question paper
10th Maths model3 question paper10th Maths model3 question paper
10th Maths model3 question paper
 
Bt0063 mathematics fot it
Bt0063 mathematics fot itBt0063 mathematics fot it
Bt0063 mathematics fot it
 
Number theory lecture (part 1)
Number theory lecture (part 1)Number theory lecture (part 1)
Number theory lecture (part 1)
 
79ecb3d9 65f4-4161-b97d-63711df5d6c5
79ecb3d9 65f4-4161-b97d-63711df5d6c579ecb3d9 65f4-4161-b97d-63711df5d6c5
79ecb3d9 65f4-4161-b97d-63711df5d6c5
 
Datamining 8th Hclustering
Datamining 8th HclusteringDatamining 8th Hclustering
Datamining 8th Hclustering
 
Datamining 8th hclustering
Datamining 8th hclusteringDatamining 8th hclustering
Datamining 8th hclustering
 
M A T H E M A T I C S I I I J N T U M O D E L P A P E R{Www
M A T H E M A T I C S  I I I  J N T U  M O D E L  P A P E R{WwwM A T H E M A T I C S  I I I  J N T U  M O D E L  P A P E R{Www
M A T H E M A T I C S I I I J N T U M O D E L P A P E R{Www
 
S101-52國立新化高中(代理)
S101-52國立新化高中(代理)S101-52國立新化高中(代理)
S101-52國立新化高中(代理)
 
Maieee03
Maieee03Maieee03
Maieee03
 
Aieee 2003 maths solved paper by fiitjee
Aieee 2003 maths solved paper by fiitjeeAieee 2003 maths solved paper by fiitjee
Aieee 2003 maths solved paper by fiitjee
 

Recently uploaded

LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
DanBrown980551
 
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
Alex Pruden
 
Dandelion Hashtable: beyond billion requests per second on a commodity server
Dandelion Hashtable: beyond billion requests per second on a commodity serverDandelion Hashtable: beyond billion requests per second on a commodity server
Dandelion Hashtable: beyond billion requests per second on a commodity server
Antonios Katsarakis
 
Christine's Supplier Sourcing Presentaion.pptx
Christine's Supplier Sourcing Presentaion.pptxChristine's Supplier Sourcing Presentaion.pptx
Christine's Supplier Sourcing Presentaion.pptx
christinelarrosa
 
A Deep Dive into ScyllaDB's Architecture
A Deep Dive into ScyllaDB's ArchitectureA Deep Dive into ScyllaDB's Architecture
A Deep Dive into ScyllaDB's Architecture
ScyllaDB
 
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeckPoznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
FilipTomaszewski5
 
Essentials of Automations: Exploring Attributes & Automation Parameters
Essentials of Automations: Exploring Attributes & Automation ParametersEssentials of Automations: Exploring Attributes & Automation Parameters
Essentials of Automations: Exploring Attributes & Automation Parameters
Safe Software
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
Chart Kalyan
 
"NATO Hackathon Winner: AI-Powered Drug Search", Taras Kloba
"NATO Hackathon Winner: AI-Powered Drug Search",  Taras Kloba"NATO Hackathon Winner: AI-Powered Drug Search",  Taras Kloba
"NATO Hackathon Winner: AI-Powered Drug Search", Taras Kloba
Fwdays
 
Biomedical Knowledge Graphs for Data Scientists and Bioinformaticians
Biomedical Knowledge Graphs for Data Scientists and BioinformaticiansBiomedical Knowledge Graphs for Data Scientists and Bioinformaticians
Biomedical Knowledge Graphs for Data Scientists and Bioinformaticians
Neo4j
 
ScyllaDB Tablets: Rethinking Replication
ScyllaDB Tablets: Rethinking ReplicationScyllaDB Tablets: Rethinking Replication
ScyllaDB Tablets: Rethinking Replication
ScyllaDB
 
Apps Break Data
Apps Break DataApps Break Data
Apps Break Data
Ivo Velitchkov
 
inQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
inQuba Webinar Mastering Customer Journey Management with Dr Graham HillinQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
inQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
LizaNolte
 
Demystifying Knowledge Management through Storytelling
Demystifying Knowledge Management through StorytellingDemystifying Knowledge Management through Storytelling
Demystifying Knowledge Management through Storytelling
Enterprise Knowledge
 
Introduction of Cybersecurity with OSS at Code Europe 2024
Introduction of Cybersecurity with OSS  at Code Europe 2024Introduction of Cybersecurity with OSS  at Code Europe 2024
Introduction of Cybersecurity with OSS at Code Europe 2024
Hiroshi SHIBATA
 
Must Know Postgres Extension for DBA and Developer during Migration
Must Know Postgres Extension for DBA and Developer during MigrationMust Know Postgres Extension for DBA and Developer during Migration
Must Know Postgres Extension for DBA and Developer during Migration
Mydbops
 
Harnessing the Power of NLP and Knowledge Graphs for Opioid Research
Harnessing the Power of NLP and Knowledge Graphs for Opioid ResearchHarnessing the Power of NLP and Knowledge Graphs for Opioid Research
Harnessing the Power of NLP and Knowledge Graphs for Opioid Research
Neo4j
 
"Frontline Battles with DDoS: Best practices and Lessons Learned", Igor Ivaniuk
"Frontline Battles with DDoS: Best practices and Lessons Learned",  Igor Ivaniuk"Frontline Battles with DDoS: Best practices and Lessons Learned",  Igor Ivaniuk
"Frontline Battles with DDoS: Best practices and Lessons Learned", Igor Ivaniuk
Fwdays
 
"What does it really mean for your system to be available, or how to define w...
"What does it really mean for your system to be available, or how to define w..."What does it really mean for your system to be available, or how to define w...
"What does it really mean for your system to be available, or how to define w...
Fwdays
 
Christine's Product Research Presentation.pptx
Christine's Product Research Presentation.pptxChristine's Product Research Presentation.pptx
Christine's Product Research Presentation.pptx
christinelarrosa
 

Recently uploaded (20)

LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
LF Energy Webinar: Carbon Data Specifications: Mechanisms to Improve Data Acc...
 
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
zkStudyClub - LatticeFold: A Lattice-based Folding Scheme and its Application...
 
Dandelion Hashtable: beyond billion requests per second on a commodity server
Dandelion Hashtable: beyond billion requests per second on a commodity serverDandelion Hashtable: beyond billion requests per second on a commodity server
Dandelion Hashtable: beyond billion requests per second on a commodity server
 
Christine's Supplier Sourcing Presentaion.pptx
Christine's Supplier Sourcing Presentaion.pptxChristine's Supplier Sourcing Presentaion.pptx
Christine's Supplier Sourcing Presentaion.pptx
 
A Deep Dive into ScyllaDB's Architecture
A Deep Dive into ScyllaDB's ArchitectureA Deep Dive into ScyllaDB's Architecture
A Deep Dive into ScyllaDB's Architecture
 
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeckPoznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
Poznań ACE event - 19.06.2024 Team 24 Wrapup slidedeck
 
Essentials of Automations: Exploring Attributes & Automation Parameters
Essentials of Automations: Exploring Attributes & Automation ParametersEssentials of Automations: Exploring Attributes & Automation Parameters
Essentials of Automations: Exploring Attributes & Automation Parameters
 
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdfHow to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
How to Interpret Trends in the Kalyan Rajdhani Mix Chart.pdf
 
"NATO Hackathon Winner: AI-Powered Drug Search", Taras Kloba
"NATO Hackathon Winner: AI-Powered Drug Search",  Taras Kloba"NATO Hackathon Winner: AI-Powered Drug Search",  Taras Kloba
"NATO Hackathon Winner: AI-Powered Drug Search", Taras Kloba
 
Biomedical Knowledge Graphs for Data Scientists and Bioinformaticians
Biomedical Knowledge Graphs for Data Scientists and BioinformaticiansBiomedical Knowledge Graphs for Data Scientists and Bioinformaticians
Biomedical Knowledge Graphs for Data Scientists and Bioinformaticians
 
ScyllaDB Tablets: Rethinking Replication
ScyllaDB Tablets: Rethinking ReplicationScyllaDB Tablets: Rethinking Replication
ScyllaDB Tablets: Rethinking Replication
 
Apps Break Data
Apps Break DataApps Break Data
Apps Break Data
 
inQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
inQuba Webinar Mastering Customer Journey Management with Dr Graham HillinQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
inQuba Webinar Mastering Customer Journey Management with Dr Graham Hill
 
Demystifying Knowledge Management through Storytelling
Demystifying Knowledge Management through StorytellingDemystifying Knowledge Management through Storytelling
Demystifying Knowledge Management through Storytelling
 
Introduction of Cybersecurity with OSS at Code Europe 2024
Introduction of Cybersecurity with OSS  at Code Europe 2024Introduction of Cybersecurity with OSS  at Code Europe 2024
Introduction of Cybersecurity with OSS at Code Europe 2024
 
Must Know Postgres Extension for DBA and Developer during Migration
Must Know Postgres Extension for DBA and Developer during MigrationMust Know Postgres Extension for DBA and Developer during Migration
Must Know Postgres Extension for DBA and Developer during Migration
 
Harnessing the Power of NLP and Knowledge Graphs for Opioid Research
Harnessing the Power of NLP and Knowledge Graphs for Opioid ResearchHarnessing the Power of NLP and Knowledge Graphs for Opioid Research
Harnessing the Power of NLP and Knowledge Graphs for Opioid Research
 
"Frontline Battles with DDoS: Best practices and Lessons Learned", Igor Ivaniuk
"Frontline Battles with DDoS: Best practices and Lessons Learned",  Igor Ivaniuk"Frontline Battles with DDoS: Best practices and Lessons Learned",  Igor Ivaniuk
"Frontline Battles with DDoS: Best practices and Lessons Learned", Igor Ivaniuk
 
"What does it really mean for your system to be available, or how to define w...
"What does it really mean for your system to be available, or how to define w..."What does it really mean for your system to be available, or how to define w...
"What does it really mean for your system to be available, or how to define w...
 
Christine's Product Research Presentation.pptx
Christine's Product Research Presentation.pptxChristine's Product Research Presentation.pptx
Christine's Product Research Presentation.pptx
 

確率伝播その2

  • 2. DAG • oscillation: • • coherent • coherent A B C E D
  • 3. 1. clustering – 2. conditioning – 3. stochastic simulation –
  • 5. A • B C Z B C z ∈ {(+b,+c), (¬b,+c), (+b,¬c), (¬b,¬c)} E D • • Z B C A P(b,c|a) Z B,C P(e|c) E D P(d|b,c)
  • 6. – (= ) – • • join trees(=clique trees or junction trees)
  • 7. Join Tree 1. NB G – – – → G NB I-map I(G)⊆I(NB) 2. G chordal supergraph G’ ※3.2.4 graph-traiangulation algorithm 3. G’ clique join tree T 4. NB – T clique-node – clique-node λ 5. T clique-node NB
  • 8. G Join Tree 1. fill-in G’ 1. maximum cardinality search • 1 |V|( ) • • 2. for (n=|V| to 1) n 2. G’ 3. C1, C2, ..., Ct 4. Ci Cj(i>j)
  • 9. Join Tree Z1 A P(d|b,c) A,B,C A P(e|c) Z2 Z3 B C B C B,C,D C,E E D λD(z2) λE(z3) D E D=0 E=1 M zi |z j = P(zi | z j ) = P(zi | zi ∩ z j ). M z2 |z1 = P(z2 | z1 ) = P(b, c, d | b, c) = P(d | b, c) M z3|z1 = P(z3 | z1 ) = P(c, e | c) = P(e | c)   0 if z2 = (b, c, +d) π (z1 ) = π z (z1 ) = π z (z1 ) = P(z1 ) = P(a, b, c) = P(b | a)P(c | a)P(a) λD (z2 ) =  2 3   1 if z2 = (b, c,¬d) π (z2 ) = ∑ P(z2 | z1 )P(z1 ) = ∑ P(d | b, c)P(a, b, c) = P(d | b, c)P(b, c)   0 if z3 = (c,¬e) Z1 a λE (z3 ) =  π (z3 ) = ∑ P(z3 | z1 )P(z1 ) = ∑ P(e | c)P(a, b, c) = P(e | c)P(c)   1 if z3 = (c,+e) Z1 a,b λZ (z1 ) = M z |z • λD (z2 ) = P(¬d | b, c) 2 2 1 BEL(z1 ) = αλZ2 (z1 )λZ3 (z1 )π (z1 ) λZ (z1 ) = M z |z • λE (z3 ) = P(+e | c) 3 3 1 = α P(¬d | b, c)P(+e | c)P(b | a)P(c | a)P(a) 8 BEL(a),BEL(b),BEL(c)
  • 10. 4.4.2 The Method of Conditioning(Reasoning by Assumptions)
  • 11. Conditioning A • – a=0 a=1 • B C • a E • BEL(b),BEL(c) D a A=0 A=0 A=1 A=1 BEL(b) = P (b | e) = ∑ P(b | a, e) P (a | e) a A A A A B C B C D=0 D E D=0 D E E=1 E=1 w0 w1 w = (w1, w0 ) = P(a) w E = P(a | +e) = α P(+e | a)P(a) w E,D = P(a | +e,¬d) = α P(¬d | a, +e)P(a | +e)
  • 12. A Short Course on Graphical Models 3 The Junction Tree Algoritms – Mark Paskin http://ai.stanaford.edu/~paskin/gm-short-course/lec3.pdf
  • 13. Join Tree? ,Junction Tree? • en.wikipedia.org/wiki/Tree_decomposition • Join Tree DAG • Junction Tree Algorithm
  • 14. Junction Tree G d T {b,d} b f {b} a e {b,e} {b,c} c {a,b,c} {b,c,e} {b,e,f} T G Junction Tree • singly connected( ): • covering: G A C A⊆C • running intersection: B C i B C i ※Junction Tree Algorithm
  • 15. Junction Tree ※ BEL λ π • • ( ) •
  • 16. message passing • message passing protocol: B C C • ( 1. Collect(C): for (B <- C.children) { Collect(B) } 2. Distribute(C): for (B <- C.children) {Distribute(B)} • The Shafer-Shenoy Algorithm: B C – C – C • {b,d} – Lauritzen-Spiegelhalter – Hugin 1{b} 4 2 1 {b,e} {b,c} {a,b,c} {b,c,e} {b,e,f} 3 4