SlideShare a Scribd company logo
Discrete MRF Inference of Marginal Densities for
Non-uniformly Discretized Variable Space
Masaki Saito, Takayuki Okatani, Koichiro Deguchi
Tohoku University
Outline
1. Introduction
2. Algorithms for a non-uniformly discretized variable space
3. Dynamic discretization method
4. Experimental results
Outline
1. Introduction
2. Algorithms for a non-uniformly discretized variable space
3. Dynamic discretization method
4. Experimental results
Markov Random Fields
MAP inference
• Directly computes the
maximum of
• Equivalent to
minimizing
MPM inference
• Estimates the marginal density
(then computes its maximum )
• Necessary in some problems
– e.g. Learning CRF models
Continuous and Discrete MRF
• Mean Field(MF) and Belief Propagation(BP) are only choices
for the estimation of marginal densities
• Two formulations: Discrete and Continuous
– The variable x at each site takes a continuous or a discrete
value
Applicability
Computational
cost (per iter.)
Discrete Wide O((# of labels)^2)
Continuous Limited Efficient
Continuous and Discrete MRF
Discretization
• A common practice is to convert a continuous problem to a
discrete one by discretizing the variable space uniformly
– A continuous density is approximated by a discrete one
Continuous and Discrete MRF
• What if the variable space can be non-uniformly discretized so
that the accuracy is maximized with a minimum number of
discretization bins?
– Will achieve a better trade-off between accuracy vs.
computational cost
– Will be useful for non-Euclidean variable spaces
uniform non-uniform Spherical surface
Outline
1. Introduction
2. Algorithms for a non-uniformly discretized variable space
3. Dynamic discretization method
4. Experimental results
• ‘s may have arbitrary positions and sizes
• ‘s may have different number of mixtures
Remark:
Our approach
1. Represent the marginal density as a mixture of rectangular
densities
2. Follow the variational method to derive MF/BP algorithms
weight rectangular density
Our approach
1. Represent the marginal density as a mixture of rectangular
densities
2. Follow the variational method to derive MF/BP algorithms
Introduce P such that
marginal densities are
easily computed
Find P minimizing
the KL distance
between P and Q
Compute the
marginal
densities
MF
[Textbook of Graphical Models by Koller-Friedman]
Conventional continuous-discrete conversion
1. Discretize the variable space into discrete labels (uniformly or
non-uniformly)
2. Define an energy E(x) in the discrete domain
 Original continuous energy (if any) is discarded
 Step 2 does not care about how the space was discretized
conventional
define a new discrete
energy function
Approximate joint
density
Models for the marginal
densities
Constraints
MF
BP
Derivation of new MF & BP algorithms
New MF and BP updating rules
• A few terms are newly added; they compensate for the non-
uniformity of descretization
– They vanish when the discretization is uniform
New updating rules Conventional updating rules
MF
BP
&
: newly added terms
Outline
1. Introduction
2. Algorithms for a non-uniformly discretized variable space
3. Dynamic discretization method
4. Experimental results
Dynamic discretization of the variable space
• If we have a prior knowledge about where is important in the
variable space, the application of our method is
straightforward
• Our “dynamic discretization method” is targeted at the case
where no such knowledge is available
1D
2D
Coarse-to-fine block subdivision
1. Discretize the variable space
coarsely (maybe uniformly)
2. Perform MF or BP iterations to
obtain the estimates of the
marginal densities
3. Identify the block whose mixture
weight is the largest, and
divide the block into subblocks
– Weights and messages are adjusted
4. Go to 2 until a desired result is
obtained
Outline
1. Introduction
2. Algorithms for a non-uniformly discretized variable space
3. Dynamic discretization method
4. Experimental results
Effect of non-uniform discretization
• A simple Gaussian grid MRF
• Non-uniform discretization; asymmetric with respect to the
origin
5
5
-1-2 210
64 16
64+16 = 80
Effect of non-uniform discretization
Mean Field
Belief Propagation
Stereo matching
• The proposed method is used to adaptively and recursively
discretize the variable space
• Conventional: divide the variable space into 16 fixed blocks
• Proposed: initially divide into 8 blocks, and finally divide into
16 blocks with 8 subdivision procedures
Conventional Proposed Input images
Ground truthx8
Stereo matching(BP)
…
Conventional
Blocks=16
Proposed
Blocks=8 Blocks=10 Blocks=16
Input images Ground truth
Blocks=12
Estimated marginal densities
Our method gives a better result than conventional one
with only 10 blocks
Disparity maps (maximum of marginal)
Conclusions
• We have derived new MF and BP algorithms for estimating
marginal densities for pairwise MRFs that can properly deal
with non-uniform discretization
• We have also presented a method that can dynamically
discretize the variable space
• These tools are expected to make it possible to achieve a
better trade-off between accuracy and computational cost
• The idea is to represent the marginal densities with a mixture
of rectangular densities and then apply the variational method
• Our approach also gives a rigorous method for converting
continuous MRFs to discrete ones

More Related Content

What's hot

ssc_icml13
ssc_icml13ssc_icml13
ssc_icml13
Guy Lebanon
 
2006 ssiai
2006 ssiai2006 ssiai
Iclr2020: Compression based bound for non-compressed network: unified general...
Iclr2020: Compression based bound for non-compressed network: unified general...Iclr2020: Compression based bound for non-compressed network: unified general...
Iclr2020: Compression based bound for non-compressed network: unified general...
Taiji Suzuki
 
Dynamic clustering algorithm using fuzzy c means
Dynamic clustering algorithm using fuzzy c meansDynamic clustering algorithm using fuzzy c means
Dynamic clustering algorithm using fuzzy c means
Wrishin Bhattacharya
 
PRML 5.5
PRML 5.5PRML 5.5
PRML 5.5
Ryuta Shitomi
 
Image Enhancement using Frequency Domain Filters
Image Enhancement using Frequency Domain FiltersImage Enhancement using Frequency Domain Filters
Image Enhancement using Frequency Domain Filters
Karthika Ramachandran
 
Digital Image Processing - Frequency Filters
Digital Image Processing - Frequency FiltersDigital Image Processing - Frequency Filters
Digital Image Processing - Frequency Filters
Aly Abdelkareem
 
Speaker Recognition using Gaussian Mixture Model
Speaker Recognition using Gaussian Mixture Model Speaker Recognition using Gaussian Mixture Model
Speaker Recognition using Gaussian Mixture Model
Saurab Dulal
 
CNN for modeling sentence
CNN for modeling sentenceCNN for modeling sentence
CNN for modeling sentence
ANISH BHANUSHALI
 
Double Slit Experiment
Double Slit ExperimentDouble Slit Experiment
Double Slit Experiment
echeng96
 
Fuzzieee-98-final
Fuzzieee-98-finalFuzzieee-98-final
Fuzzieee-98-final
Sumit Sen
 
Reconstruction of partial sum without gibbs phenomenon
Reconstruction of partial sum without gibbs phenomenonReconstruction of partial sum without gibbs phenomenon
Reconstruction of partial sum without gibbs phenomenon
Deahyun Kim
 
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
Wuhyun Rico Shin
 
Frequency Image Processing
Frequency Image ProcessingFrequency Image Processing
Frequency Image Processing
Suhas Deshpande
 
Id135
Id135Id135
Id135
IJEEE
 
Machine Learning: An Introduction Fu Chang
Machine Learning: An Introduction Fu ChangMachine Learning: An Introduction Fu Chang
Machine Learning: An Introduction Fu Chang
butest
 
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
Edureka!
 
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
Taiji Suzuki
 
Ch7 noise variation of different modulation scheme pg 63
Ch7 noise variation of different modulation scheme pg 63Ch7 noise variation of different modulation scheme pg 63
Ch7 noise variation of different modulation scheme pg 63
Prateek Omer
 
Design of Low-Pass Digital Differentiators Based on B-splines
Design of Low-Pass Digital Differentiators Based on B-splinesDesign of Low-Pass Digital Differentiators Based on B-splines
Design of Low-Pass Digital Differentiators Based on B-splines
CSCJournals
 

What's hot (20)

ssc_icml13
ssc_icml13ssc_icml13
ssc_icml13
 
2006 ssiai
2006 ssiai2006 ssiai
2006 ssiai
 
Iclr2020: Compression based bound for non-compressed network: unified general...
Iclr2020: Compression based bound for non-compressed network: unified general...Iclr2020: Compression based bound for non-compressed network: unified general...
Iclr2020: Compression based bound for non-compressed network: unified general...
 
Dynamic clustering algorithm using fuzzy c means
Dynamic clustering algorithm using fuzzy c meansDynamic clustering algorithm using fuzzy c means
Dynamic clustering algorithm using fuzzy c means
 
PRML 5.5
PRML 5.5PRML 5.5
PRML 5.5
 
Image Enhancement using Frequency Domain Filters
Image Enhancement using Frequency Domain FiltersImage Enhancement using Frequency Domain Filters
Image Enhancement using Frequency Domain Filters
 
Digital Image Processing - Frequency Filters
Digital Image Processing - Frequency FiltersDigital Image Processing - Frequency Filters
Digital Image Processing - Frequency Filters
 
Speaker Recognition using Gaussian Mixture Model
Speaker Recognition using Gaussian Mixture Model Speaker Recognition using Gaussian Mixture Model
Speaker Recognition using Gaussian Mixture Model
 
CNN for modeling sentence
CNN for modeling sentenceCNN for modeling sentence
CNN for modeling sentence
 
Double Slit Experiment
Double Slit ExperimentDouble Slit Experiment
Double Slit Experiment
 
Fuzzieee-98-final
Fuzzieee-98-finalFuzzieee-98-final
Fuzzieee-98-final
 
Reconstruction of partial sum without gibbs phenomenon
Reconstruction of partial sum without gibbs phenomenonReconstruction of partial sum without gibbs phenomenon
Reconstruction of partial sum without gibbs phenomenon
 
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
Paper review: Measuring the Intrinsic Dimension of Objective Landscapes.
 
Frequency Image Processing
Frequency Image ProcessingFrequency Image Processing
Frequency Image Processing
 
Id135
Id135Id135
Id135
 
Machine Learning: An Introduction Fu Chang
Machine Learning: An Introduction Fu ChangMachine Learning: An Introduction Fu Chang
Machine Learning: An Introduction Fu Chang
 
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
Restricted Boltzmann Machine | Neural Network Tutorial | Deep Learning Tutori...
 
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
[NeurIPS2020 (spotlight)] Generalization bound of globally optimal non convex...
 
Ch7 noise variation of different modulation scheme pg 63
Ch7 noise variation of different modulation scheme pg 63Ch7 noise variation of different modulation scheme pg 63
Ch7 noise variation of different modulation scheme pg 63
 
Design of Low-Pass Digital Differentiators Based on B-splines
Design of Low-Pass Digital Differentiators Based on B-splinesDesign of Low-Pass Digital Differentiators Based on B-splines
Design of Low-Pass Digital Differentiators Based on B-splines
 

Viewers also liked

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
 
Python で画像処理をやってみよう! -SIFT 第7回-
Python で画像処理をやってみよう! -SIFT 第7回-Python で画像処理をやってみよう! -SIFT 第7回-
Python で画像処理をやってみよう! -SIFT 第7回-
Project Samurai
 
グラフィカルモデル入門
グラフィカルモデル入門グラフィカルモデル入門
グラフィカルモデル入門Kawamoto_Kazuhiko
 
1章前半
1章前半1章前半
1章前半
Ogawa Shuhei
 
3Com 3C6555-016R EFSM
3Com 3C6555-016R  EFSM3Com 3C6555-016R  EFSM
3Com 3C6555-016R EFSM
savomir
 
第4回MachineLearningのための数学塾資料(浅川)
第4回MachineLearningのための数学塾資料(浅川)第4回MachineLearningのための数学塾資料(浅川)
第4回MachineLearningのための数学塾資料(浅川)
Shin Asakawa
 
chapter 11 structural svm
chapter 11 structural svmchapter 11 structural svm
chapter 11 structural svm
Taikai Takeda
 
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...takaya imai
 
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
Project Samurai
 
闇と向き合う
闇と向き合う闇と向き合う
闇と向き合う
Nagi Teramo
 
TensorFlow math ja 05 word2vec
TensorFlow math ja 05 word2vecTensorFlow math ja 05 word2vec
TensorFlow math ja 05 word2vec
Shin Asakawa
 
動的輪郭モデル
動的輪郭モデル動的輪郭モデル
動的輪郭モデル
Arumaziro
 
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
Takuma Yagi
 
拡張現実に必要な技術
拡張現実に必要な技術拡張現実に必要な技術
拡張現実に必要な技術
Ryo Nakamura
 
20160825 IEICE SIP研究会 講演
20160825 IEICE SIP研究会 講演20160825 IEICE SIP研究会 講演
20160825 IEICE SIP研究会 講演
Hayaru SHOUNO
 
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
Project Samurai
 
深層学習(ディープラーニング)入門勉強会資料(浅川)
深層学習(ディープラーニング)入門勉強会資料(浅川)深層学習(ディープラーニング)入門勉強会資料(浅川)
深層学習(ディープラーニング)入門勉強会資料(浅川)
Shin Asakawa
 
ICML読み会2016@早稲田
ICML読み会2016@早稲田ICML読み会2016@早稲田
ICML読み会2016@早稲田
Taikai Takeda
 
healthplanetパッケージで 体組成データを手に入れて 健康な体も手に入れる
healthplanetパッケージで体組成データを手に入れて健康な体も手に入れるhealthplanetパッケージで体組成データを手に入れて健康な体も手に入れる
healthplanetパッケージで 体組成データを手に入れて 健康な体も手に入れる
Nagi Teramo
 
MLP SVM Chapter 7 分割法
MLP SVM Chapter 7 分割法MLP SVM Chapter 7 分割法
MLP SVM Chapter 7 分割法
Taikai Takeda
 

Viewers also liked (20)

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
 
Python で画像処理をやってみよう! -SIFT 第7回-
Python で画像処理をやってみよう! -SIFT 第7回-Python で画像処理をやってみよう! -SIFT 第7回-
Python で画像処理をやってみよう! -SIFT 第7回-
 
グラフィカルモデル入門
グラフィカルモデル入門グラフィカルモデル入門
グラフィカルモデル入門
 
1章前半
1章前半1章前半
1章前半
 
3Com 3C6555-016R EFSM
3Com 3C6555-016R  EFSM3Com 3C6555-016R  EFSM
3Com 3C6555-016R EFSM
 
第4回MachineLearningのための数学塾資料(浅川)
第4回MachineLearningのための数学塾資料(浅川)第4回MachineLearningのための数学塾資料(浅川)
第4回MachineLearningのための数学塾資料(浅川)
 
chapter 11 structural svm
chapter 11 structural svmchapter 11 structural svm
chapter 11 structural svm
 
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...
New Approach to Data Analysis System “Tableau + TreasureData” at Tableau User...
 
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
数学的基礎から学ぶ Deep Learning (with Python) Vol. 9
 
闇と向き合う
闇と向き合う闇と向き合う
闇と向き合う
 
TensorFlow math ja 05 word2vec
TensorFlow math ja 05 word2vecTensorFlow math ja 05 word2vec
TensorFlow math ja 05 word2vec
 
動的輪郭モデル
動的輪郭モデル動的輪郭モデル
動的輪郭モデル
 
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
論文紹介:Using the Forest to See the Trees: A Graphical. Model Relating Features,...
 
拡張現実に必要な技術
拡張現実に必要な技術拡張現実に必要な技術
拡張現実に必要な技術
 
20160825 IEICE SIP研究会 講演
20160825 IEICE SIP研究会 講演20160825 IEICE SIP研究会 講演
20160825 IEICE SIP研究会 講演
 
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
数学的基礎から学ぶ Deep Learning (with Python) Vol. 12
 
深層学習(ディープラーニング)入門勉強会資料(浅川)
深層学習(ディープラーニング)入門勉強会資料(浅川)深層学習(ディープラーニング)入門勉強会資料(浅川)
深層学習(ディープラーニング)入門勉強会資料(浅川)
 
ICML読み会2016@早稲田
ICML読み会2016@早稲田ICML読み会2016@早稲田
ICML読み会2016@早稲田
 
healthplanetパッケージで 体組成データを手に入れて 健康な体も手に入れる
healthplanetパッケージで体組成データを手に入れて健康な体も手に入れるhealthplanetパッケージで体組成データを手に入れて健康な体も手に入れる
healthplanetパッケージで 体組成データを手に入れて 健康な体も手に入れる
 
MLP SVM Chapter 7 分割法
MLP SVM Chapter 7 分割法MLP SVM Chapter 7 分割法
MLP SVM Chapter 7 分割法
 

Similar to Discrete MRF Inference of Marginal Densities for Non-uniformly Discretized Variable Space

STAN_MS_PPT.pptx
STAN_MS_PPT.pptxSTAN_MS_PPT.pptx
STAN_MS_PPT.pptx
Md Abul Hayat
 
머피의 머신러닝: 17장 Markov Chain and HMM
머피의 머신러닝: 17장  Markov Chain and HMM머피의 머신러닝: 17장  Markov Chain and HMM
머피의 머신러닝: 17장 Markov Chain and HMM
Jungkyu Lee
 
Markov Chain Monte Carlo explained
Markov Chain Monte Carlo explainedMarkov Chain Monte Carlo explained
Markov Chain Monte Carlo explained
dariodigiuni
 
IMPT and optimization
IMPT and optimizationIMPT and optimization
IMPT and optimization
Veronika Paštyková
 
Firefly exact MCMC for Big Data
Firefly exact MCMC for Big DataFirefly exact MCMC for Big Data
Firefly exact MCMC for Big Data
Gianvito Siciliano
 
第13回 配信講義 計算科学技術特論A(2021)
第13回 配信講義 計算科学技術特論A(2021)第13回 配信講義 計算科学技術特論A(2021)
第13回 配信講義 計算科学技術特論A(2021)
RCCSRENKEI
 
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
The Statistical and Applied Mathematical Sciences Institute
 
Efficient methods for accurately calculating thermoelectric properties – elec...
Efficient methods for accurately calculating thermoelectric properties – elec...Efficient methods for accurately calculating thermoelectric properties – elec...
Efficient methods for accurately calculating thermoelectric properties – elec...
Anubhav Jain
 
Semi orthogonal low-rank matrix factorization for deep neural networks
Semi orthogonal low-rank matrix factorization for deep neural networksSemi orthogonal low-rank matrix factorization for deep neural networks
Semi orthogonal low-rank matrix factorization for deep neural networks
品媛 陳
 
various methods for image segmentation
various methods for image segmentationvarious methods for image segmentation
various methods for image segmentation
Raveesh Methi
 
Lecture3
Lecture3Lecture3
Lecture3
Heather Kulik
 
CSA 3702 machine learning module 3
CSA 3702 machine learning module 3CSA 3702 machine learning module 3
CSA 3702 machine learning module 3
Nandhini S
 
Introduction to finite element method
Introduction to finite element methodIntroduction to finite element method
Introduction to finite element method
shahzaib601980
 
generalized_nbody_acs_2015_challacombe
generalized_nbody_acs_2015_challacombegeneralized_nbody_acs_2015_challacombe
generalized_nbody_acs_2015_challacombe
Matt Challacombe
 
El text.tokuron a(2019).jung190711
El text.tokuron a(2019).jung190711El text.tokuron a(2019).jung190711
El text.tokuron a(2019).jung190711
RCCSRENKEI
 
Ec 2401 wireless communication unit 4
Ec 2401 wireless communication   unit 4Ec 2401 wireless communication   unit 4
Ec 2401 wireless communication unit 4
JAIGANESH SEKAR
 
Oceans13 Presentation
Oceans13 PresentationOceans13 Presentation
Oceans13 Presentation
Ahmad ElMoslimany
 
12 l1-harmonic methodology
12 l1-harmonic methodology12 l1-harmonic methodology
12 l1-harmonic methodology
chowdavaramsaiprasad
 
Lab4 slides
Lab4 slidesLab4 slides
Lab4 slides
Heather Kulik
 
Second or fourth-order finite difference operators, which one is most effective?
Second or fourth-order finite difference operators, which one is most effective?Second or fourth-order finite difference operators, which one is most effective?
Second or fourth-order finite difference operators, which one is most effective?
Premier Publishers
 

Similar to Discrete MRF Inference of Marginal Densities for Non-uniformly Discretized Variable Space (20)

STAN_MS_PPT.pptx
STAN_MS_PPT.pptxSTAN_MS_PPT.pptx
STAN_MS_PPT.pptx
 
머피의 머신러닝: 17장 Markov Chain and HMM
머피의 머신러닝: 17장  Markov Chain and HMM머피의 머신러닝: 17장  Markov Chain and HMM
머피의 머신러닝: 17장 Markov Chain and HMM
 
Markov Chain Monte Carlo explained
Markov Chain Monte Carlo explainedMarkov Chain Monte Carlo explained
Markov Chain Monte Carlo explained
 
IMPT and optimization
IMPT and optimizationIMPT and optimization
IMPT and optimization
 
Firefly exact MCMC for Big Data
Firefly exact MCMC for Big DataFirefly exact MCMC for Big Data
Firefly exact MCMC for Big Data
 
第13回 配信講義 計算科学技術特論A(2021)
第13回 配信講義 計算科学技術特論A(2021)第13回 配信講義 計算科学技術特論A(2021)
第13回 配信講義 計算科学技術特論A(2021)
 
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
Program on Quasi-Monte Carlo and High-Dimensional Sampling Methods for Applie...
 
Efficient methods for accurately calculating thermoelectric properties – elec...
Efficient methods for accurately calculating thermoelectric properties – elec...Efficient methods for accurately calculating thermoelectric properties – elec...
Efficient methods for accurately calculating thermoelectric properties – elec...
 
Semi orthogonal low-rank matrix factorization for deep neural networks
Semi orthogonal low-rank matrix factorization for deep neural networksSemi orthogonal low-rank matrix factorization for deep neural networks
Semi orthogonal low-rank matrix factorization for deep neural networks
 
various methods for image segmentation
various methods for image segmentationvarious methods for image segmentation
various methods for image segmentation
 
Lecture3
Lecture3Lecture3
Lecture3
 
CSA 3702 machine learning module 3
CSA 3702 machine learning module 3CSA 3702 machine learning module 3
CSA 3702 machine learning module 3
 
Introduction to finite element method
Introduction to finite element methodIntroduction to finite element method
Introduction to finite element method
 
generalized_nbody_acs_2015_challacombe
generalized_nbody_acs_2015_challacombegeneralized_nbody_acs_2015_challacombe
generalized_nbody_acs_2015_challacombe
 
El text.tokuron a(2019).jung190711
El text.tokuron a(2019).jung190711El text.tokuron a(2019).jung190711
El text.tokuron a(2019).jung190711
 
Ec 2401 wireless communication unit 4
Ec 2401 wireless communication   unit 4Ec 2401 wireless communication   unit 4
Ec 2401 wireless communication unit 4
 
Oceans13 Presentation
Oceans13 PresentationOceans13 Presentation
Oceans13 Presentation
 
12 l1-harmonic methodology
12 l1-harmonic methodology12 l1-harmonic methodology
12 l1-harmonic methodology
 
Lab4 slides
Lab4 slidesLab4 slides
Lab4 slides
 
Second or fourth-order finite difference operators, which one is most effective?
Second or fourth-order finite difference operators, which one is most effective?Second or fourth-order finite difference operators, which one is most effective?
Second or fourth-order finite difference operators, which one is most effective?
 

Recently uploaded

“I’m still / I’m still / Chaining from the Block”
“I’m still / I’m still / Chaining from the Block”“I’m still / I’m still / Chaining from the Block”
“I’m still / I’m still / Chaining from the Block”
Claudio Di Ciccio
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
mikeeftimakis1
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Malak Abu Hammad
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
Safe Software
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Paige Cruz
 
GenAI Pilot Implementation in the organizations
GenAI Pilot Implementation in the organizationsGenAI Pilot Implementation in the organizations
GenAI Pilot Implementation in the organizations
kumardaparthi1024
 
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with SlackLet's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
shyamraj55
 
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAUHCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
panagenda
 
RESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for studentsRESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for students
KAMESHS29
 
HCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAUHCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAU
panagenda
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
Kumud Singh
 
How to use Firebase Data Connect For Flutter
How to use Firebase Data Connect For FlutterHow to use Firebase Data Connect For Flutter
How to use Firebase Data Connect For Flutter
Daiki Mogmet Ito
 
Presentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of GermanyPresentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of Germany
innovationoecd
 
Artificial Intelligence for XMLDevelopment
Artificial Intelligence for XMLDevelopmentArtificial Intelligence for XMLDevelopment
Artificial Intelligence for XMLDevelopment
Octavian Nadolu
 
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
Edge AI and Vision Alliance
 
Video Streaming: Then, Now, and in the Future
Video Streaming: Then, Now, and in the FutureVideo Streaming: Then, Now, and in the Future
Video Streaming: Then, Now, and in the Future
Alpen-Adria-Universität
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
Adtran
 
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
Neo4j
 
UiPath Test Automation using UiPath Test Suite series, part 6
UiPath Test Automation using UiPath Test Suite series, part 6UiPath Test Automation using UiPath Test Suite series, part 6
UiPath Test Automation using UiPath Test Suite series, part 6
DianaGray10
 
20240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 202420240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 2024
Matthew Sinclair
 

Recently uploaded (20)

“I’m still / I’m still / Chaining from the Block”
“I’m still / I’m still / Chaining from the Block”“I’m still / I’m still / Chaining from the Block”
“I’m still / I’m still / Chaining from the Block”
 
Introduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - CybersecurityIntroduction to CHERI technology - Cybersecurity
Introduction to CHERI technology - Cybersecurity
 
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdfUnlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
Unlock the Future of Search with MongoDB Atlas_ Vector Search Unleashed.pdf
 
Essentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FMEEssentials of Automations: The Art of Triggers and Actions in FME
Essentials of Automations: The Art of Triggers and Actions in FME
 
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdfObservability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
Observability Concepts EVERY Developer Should Know -- DeveloperWeek Europe.pdf
 
GenAI Pilot Implementation in the organizations
GenAI Pilot Implementation in the organizationsGenAI Pilot Implementation in the organizations
GenAI Pilot Implementation in the organizations
 
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with SlackLet's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
Let's Integrate MuleSoft RPA, COMPOSER, APM with AWS IDP along with Slack
 
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAUHCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
HCL Notes und Domino Lizenzkostenreduzierung in der Welt von DLAU
 
RESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for studentsRESUME BUILDER APPLICATION Project for students
RESUME BUILDER APPLICATION Project for students
 
HCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAUHCL Notes and Domino License Cost Reduction in the World of DLAU
HCL Notes and Domino License Cost Reduction in the World of DLAU
 
Mind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AIMind map of terminologies used in context of Generative AI
Mind map of terminologies used in context of Generative AI
 
How to use Firebase Data Connect For Flutter
How to use Firebase Data Connect For FlutterHow to use Firebase Data Connect For Flutter
How to use Firebase Data Connect For Flutter
 
Presentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of GermanyPresentation of the OECD Artificial Intelligence Review of Germany
Presentation of the OECD Artificial Intelligence Review of Germany
 
Artificial Intelligence for XMLDevelopment
Artificial Intelligence for XMLDevelopmentArtificial Intelligence for XMLDevelopment
Artificial Intelligence for XMLDevelopment
 
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
“Building and Scaling AI Applications with the Nx AI Manager,” a Presentation...
 
Video Streaming: Then, Now, and in the Future
Video Streaming: Then, Now, and in the FutureVideo Streaming: Then, Now, and in the Future
Video Streaming: Then, Now, and in the Future
 
Pushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 daysPushing the limits of ePRTC: 100ns holdover for 100 days
Pushing the limits of ePRTC: 100ns holdover for 100 days
 
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
GraphSummit Singapore | Graphing Success: Revolutionising Organisational Stru...
 
UiPath Test Automation using UiPath Test Suite series, part 6
UiPath Test Automation using UiPath Test Suite series, part 6UiPath Test Automation using UiPath Test Suite series, part 6
UiPath Test Automation using UiPath Test Suite series, part 6
 
20240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 202420240607 QFM018 Elixir Reading List May 2024
20240607 QFM018 Elixir Reading List May 2024
 

Discrete MRF Inference of Marginal Densities for Non-uniformly Discretized Variable Space

  • 1. Discrete MRF Inference of Marginal Densities for Non-uniformly Discretized Variable Space Masaki Saito, Takayuki Okatani, Koichiro Deguchi Tohoku University
  • 2. Outline 1. Introduction 2. Algorithms for a non-uniformly discretized variable space 3. Dynamic discretization method 4. Experimental results
  • 3. Outline 1. Introduction 2. Algorithms for a non-uniformly discretized variable space 3. Dynamic discretization method 4. Experimental results
  • 4. Markov Random Fields MAP inference • Directly computes the maximum of • Equivalent to minimizing MPM inference • Estimates the marginal density (then computes its maximum ) • Necessary in some problems – e.g. Learning CRF models
  • 5. Continuous and Discrete MRF • Mean Field(MF) and Belief Propagation(BP) are only choices for the estimation of marginal densities • Two formulations: Discrete and Continuous – The variable x at each site takes a continuous or a discrete value Applicability Computational cost (per iter.) Discrete Wide O((# of labels)^2) Continuous Limited Efficient
  • 6. Continuous and Discrete MRF Discretization • A common practice is to convert a continuous problem to a discrete one by discretizing the variable space uniformly – A continuous density is approximated by a discrete one
  • 7. Continuous and Discrete MRF • What if the variable space can be non-uniformly discretized so that the accuracy is maximized with a minimum number of discretization bins? – Will achieve a better trade-off between accuracy vs. computational cost – Will be useful for non-Euclidean variable spaces uniform non-uniform Spherical surface
  • 8. Outline 1. Introduction 2. Algorithms for a non-uniformly discretized variable space 3. Dynamic discretization method 4. Experimental results
  • 9. • ‘s may have arbitrary positions and sizes • ‘s may have different number of mixtures Remark: Our approach 1. Represent the marginal density as a mixture of rectangular densities 2. Follow the variational method to derive MF/BP algorithms weight rectangular density
  • 10. Our approach 1. Represent the marginal density as a mixture of rectangular densities 2. Follow the variational method to derive MF/BP algorithms Introduce P such that marginal densities are easily computed Find P minimizing the KL distance between P and Q Compute the marginal densities MF [Textbook of Graphical Models by Koller-Friedman]
  • 11. Conventional continuous-discrete conversion 1. Discretize the variable space into discrete labels (uniformly or non-uniformly) 2. Define an energy E(x) in the discrete domain  Original continuous energy (if any) is discarded  Step 2 does not care about how the space was discretized conventional define a new discrete energy function
  • 12. Approximate joint density Models for the marginal densities Constraints MF BP Derivation of new MF & BP algorithms
  • 13. New MF and BP updating rules • A few terms are newly added; they compensate for the non- uniformity of descretization – They vanish when the discretization is uniform New updating rules Conventional updating rules MF BP & : newly added terms
  • 14. Outline 1. Introduction 2. Algorithms for a non-uniformly discretized variable space 3. Dynamic discretization method 4. Experimental results
  • 15. Dynamic discretization of the variable space • If we have a prior knowledge about where is important in the variable space, the application of our method is straightforward • Our “dynamic discretization method” is targeted at the case where no such knowledge is available 1D 2D
  • 16. Coarse-to-fine block subdivision 1. Discretize the variable space coarsely (maybe uniformly) 2. Perform MF or BP iterations to obtain the estimates of the marginal densities 3. Identify the block whose mixture weight is the largest, and divide the block into subblocks – Weights and messages are adjusted 4. Go to 2 until a desired result is obtained
  • 17. Outline 1. Introduction 2. Algorithms for a non-uniformly discretized variable space 3. Dynamic discretization method 4. Experimental results
  • 18. Effect of non-uniform discretization • A simple Gaussian grid MRF • Non-uniform discretization; asymmetric with respect to the origin 5 5 -1-2 210 64 16 64+16 = 80
  • 19. Effect of non-uniform discretization Mean Field Belief Propagation
  • 20. Stereo matching • The proposed method is used to adaptively and recursively discretize the variable space • Conventional: divide the variable space into 16 fixed blocks • Proposed: initially divide into 8 blocks, and finally divide into 16 blocks with 8 subdivision procedures Conventional Proposed Input images Ground truthx8
  • 21. Stereo matching(BP) … Conventional Blocks=16 Proposed Blocks=8 Blocks=10 Blocks=16 Input images Ground truth Blocks=12 Estimated marginal densities Our method gives a better result than conventional one with only 10 blocks Disparity maps (maximum of marginal)
  • 22. Conclusions • We have derived new MF and BP algorithms for estimating marginal densities for pairwise MRFs that can properly deal with non-uniform discretization • We have also presented a method that can dynamically discretize the variable space • These tools are expected to make it possible to achieve a better trade-off between accuracy and computational cost • The idea is to represent the marginal densities with a mixture of rectangular densities and then apply the variational method • Our approach also gives a rigorous method for converting continuous MRFs to discrete ones