Download presentation
Presentation is loading. Please wait.
Published byCandice Richardson Modified over 9 years ago
1
September 2007 IW-SMI2007, Kyoto 1 A Quantum-Statistical-Mechanical Extension of Gaussian Mixture Model Kazuyuki Tanaka Graduate School of Information Sciences, Tohoku University, Sendai, Japan http://www.smapip.is.tohoku.ac.jp/~kazu/ In collaboration with Koji Tsuda Max Planck Institute for Biological Cybernetics, Germany
2
September 2007IW-SMI2007, Kyoto2 Contents 1.Introduction 2.Conventional Gaussian Mixture Model 3.Quantum Mechanical Extension of Gaussian Mixture Model 4.Quantum Belief Propagation 5.Concluding Remarks
3
September 2007IW-SMI2007, Kyoto3 Information Processing by using Quantum Statistical-Mechanics Quantum Annealing in Optimizations Quantum Error Correcting Codes etc... Massive Information Processing by means of Density Matrix
4
September 2007IW-SMI2007, Kyoto4 Motivations How can we construct the quantum Gaussian mixture model? How can we construct a data- classification algorithm by using the quantum Gaussian mixture model?
5
September 2007IW-SMI2007, Kyoto5 Contents 1.Introduction 2.Conventional Gaussian Mixture Model 3.Quantum Mechanical Extension of Gaussian Mixture Model 4.Quantum Belief Propagation 5.Concluding Remarks
6
September 2007IW-SMI2007, Kyoto6 Prior of Gauss Mixture Model 1 32 Histogram Label x i is generated randomly and independently of each node. 3 labels x i =1 x i =2x i =3 One of three labels 1,2 and 3 is assigned to each node.
7
September 2007IW-SMI2007, Kyoto7 Date Generating Process Data y i are generated randomly and independently of each node. x i =1 x i =2 x i =3
8
September 2007IW-SMI2007, Kyoto8 Gauss Mixture Models Prior Probability Data Generating Process Marginal Likelihood for Hyperparameters , and
9
September 2007IW-SMI2007, Kyoto9 Conventional Gauss Mixture Models , , (yi)(yi) Labels
10
September 2007IW-SMI2007, Kyoto10 Contents 1.Introduction 2.Conventional Gaussian Mixture Model 3.Quantum Mechanical Extension of Gaussian Mixture Model 4.Quantum Belief Propagation 5.Concluding Remarks
11
September 2007IW-SMI2007, Kyoto11 Quantum Gauss Mixture Models
12
September 2007IW-SMI2007, Kyoto12 Quantum Gauss Mixture Models Quantum Representation
13
September 2007IW-SMI2007, Kyoto13 Quantum Gauss Mixture Models
14
September 2007IW-SMI2007, Kyoto14 Linear Response Formulas Maxmum Likelihood Estimation in Quantum Gauss Mixture Model
15
September 2007IW-SMI2007, Kyoto15 Quantum Gauss Mixture Models , , (yi)(yi)
16
September 2007IW-SMI2007, Kyoto16 Image Segmentation Original Image Histogram Conventional Gauss Mixture Model Quantum Gauss Mixture Model = 0.2 = 0.4 0 255 0 0
17
September 2007IW-SMI2007, Kyoto17 Image Segmentation Original Image Histogram Conventional Gauss Mixture Model Quantum Gauss Mixture Model = 0.5 = 1.0 02550 0
18
September 2007IW-SMI2007, Kyoto18 Contents 1.Introduction 2.Conventional Gaussian Mixture Model 3.Quantum Mechanical Extension of Gaussian Mixture Model 4.Quantum Belief Propagation 5.Concluding Remarks
19
September 2007 IW-SMI2007, Kyoto 19 Image Segmentation by Combining Gauss Mixture Model with Potts Model Belief Propagation = = > Potts Model 4 labels
20
September 2007IW-SMI2007, Kyoto20 Image Segmentation Original Image Histogram Gauss Mixture Model Gauss Mixture Model and Potts Model BeliefPropagation
21
September 2007IW-SMI2007, Kyoto21 Density Matrix and Reduced Density Matrix Reduced Density Matrix Reducibility Condition j i
22
September 2007IW-SMI2007, Kyoto22 Reduced Density Matrix and Effective Fields i j i All effective field are matrices
23
September 2007IW-SMI2007, Kyoto23 Belief Propagation for Quantum Statistical Systems Propagating Rule of Effective Fields j i Output
24
September 2007IW-SMI2007, Kyoto24 Contents 1.Introduction 2.Conventional Gaussian Mixture Model 3.Quantum Mechanical Extension of Gaussian Mixture Model 4.Quantum Belief Propagation 5.Concluding Remarks
25
September 2007IW-SMI2007, Kyoto25 Summary An Extension to Quantum Statistical Mechanical Gaussian Mixture Model Practical Algorithm Linear Response Formula Application of Potts Model and Quantum Belief Propagation Applications to Data Mining Extension to Quantum Deterministic Annealing Future Problem
Similar presentations
© 2024 SlidePlayer.com. Inc.
All rights reserved.