neurally guided structure inference
play

Neurally-Guided Structure Inference http://ngsi.csail.mit.edu Sidi - PowerPoint PPT Presentation

Jan 30, 2024 •524 likes •925 views

Neurally-Guided Structure Inference http://ngsi.csail.mit.edu Sidi Lu*, Jiayuan Mao*, Josh Tenenbaum, and Jiajun Wu (* indicates equal contributions) Structure Inference Structure Inference Data [Kemp et al. 2008] Structure Inference Data

  1. Neurally-Guided Structure Inference http://ngsi.csail.mit.edu Sidi Lu*, Jiayuan Mao*, Josh Tenenbaum, and Jiajun Wu (* indicates equal contributions)

  2. Structure Inference

  3. Structure Inference Data [Kemp et al. 2008]

  4. Structure Inference Data Structure [Kemp et al. 2008]

  5. Structure Inference Data Structure [Kemp et al. 2008] [LeCun et al. 1998]

  6. Structure Inference Data Structure Inference Structure [Kemp et al. 2008] [LeCun et al. 1998] [Chen et al. 2016]

  7. Matrix Decomposition Models • Clustering MG+G • Low-Rank Approximation GG+G • Binary Features BG+G • Random Walk CG+G • Co-Clustering M(GM T +G)+G • Clustered Matrix Decomp. (MG+G)(GM T +G)+G • Binary Matrix Factorization (BG+G)(GB T +G)+G • Dependent GSM (exp(GG+G)◦G)+G • ……

  8. Structure Inference with Matrix Decomposition Models [Grosse et al. 2012]

  9. Structure Inference with Matrix Decomposition Models −1.01 1.36 −1.64 −0.76 0.48 −0.50 ⋮ ⋮ ⋮ −0.24 0.89 −1.12 Input Matrix 20×3 [Grosse et al. 2012]

  10. Structure Inference with Matrix Decomposition Models −1.01 1.36 −1.64 −0.76 0.48 −0.50 ⋮ ⋮ ⋮ −0.24 0.89 −1.12 Input Matrix 20×3 Structure ! [Grosse et al. 2012]

  11. Structure Inference with Matrix Decomposition Models −1.01 1.36 −1.64 −0.76 0.48 −0.50 ⋮ ⋮ ⋮ −0.24 0.89 −1.12 Input Matrix 20×3 Structure ! [Grosse et al. 2012]

  12. Structure Inference with Matrix Decomposition Models −1.01 1.36 −1.64 Cluster −0.76 0.48 −0.50 ⋮ ⋮ ⋮ −0.24 0.89 −1.12 Input Matrix 20×3 Structure ! [Grosse et al. 2012]

  13. Structure Inference with Matrix Decomposition Models 0 0 0 0 1 −1.01 1.36 −1.64 Cluster 0 0 0 1 0 −0.76 0.48 −0.50 ⋮ ⋮ ⋮ ⋮ −0.24 0.89 −1.12 0 0 1 0 0 Input Matrix Cluster Label 20×3 20×5 Structure ! [Grosse et al. 2012]

  14. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center 20×3 20×5 5× 3 Structure ! [Grosse et al. 2012]

  15. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! [Grosse et al. 2012]

  16. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! [Grosse et al. 2012]

  17. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! [Grosse et al. 2012]

  18. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! LowRank [Grosse et al. 2012]

  19. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! −1.32 +2.01 0 0 0 0 1 −0.24 −0.04 0.01 LowRank +0.04 −0.23 0 0 0 1 0 1 0 −0.22 −0.61 0.76 −04 + +0.08 +0.95 ⋮ ⋮ ⋮ ⋮ 0 1 −1.30 −0.68 +0.52 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 Cluster Noise Cluster Label Cluster Center 20×5 5×2 @(2×3) 20× 3 [Grosse et al. 2012]

  20. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure ! Structure 1! + ! −1.32 +2.01 0 0 0 0 1 −0.24 −0.04 0.01 LowRank +0.04 −0.23 0 0 0 1 0 1 0 −0.22 −0.61 0.76 −04 + +0.08 +0.95 ⋮ ⋮ ⋮ ⋮ 0 1 −1.30 −0.68 +0.52 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 Cluster Noise Cluster Label Cluster Center 20×5 5×2 @(2×3) 20× 3 Structure 1(!! + !) + ! [Grosse et al. 2012]

  21. Structure Inference with Matrix Decomposition Models −1.32 +2.01 −2.33 0 0 0 0 1 −1.01 1.36 −1.64 −0.24 −0.04 0.01 Cluster +0.04 −0.23 −0.30 0 0 0 1 0 −0.76 0.48 −0.50 −0.61 0.76 −04 + +0.08 +0.95 −1.27 ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ −0.68 +0.52 −0.54 −0.24 0.89 −1.12 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 −1.66 Input Matrix Cluster Label Cluster Center Cluster Noise 20×3 20×5 5× 3 20×3 Structure 0 Structure 10 + 0 −1.32 +2.01 0 0 0 0 1 −0.24 −0.04 0.01 LowRank +0.04 −0.23 0 0 0 1 0 1 0 −0.22 −0.61 0.76 −04 + +0.08 +0.95 ⋮ ⋮ ⋮ ⋮ 0 1 −1.30 −0.68 +0.52 −0.09 0.90 −1.84 0 0 1 0 0 −0.77 +1.40 Cluster Noise Cluster Label Cluster Center 20×5 5×2 @(2×3) 20× 3 Structure 1(00 + 0) + 0 [Grosse et al. 2012]

  22. Naïve Exhaustive Search • Clustering MG+G • Low-Rank Approximation GG+G • Binary Features BG+G • Random Walk CG+G • Co-Clustering M(GM T +G)+G • Clustered Matrix Decomp. (MG+G)(GM T +G)+G • Binary Matrix Factorization (BG+G)(GB T +G)+G • Dependent GSM (exp(GG+G)◦G)+G • ……

  23. Naïve Exhaustive Search • Clustering MG+G Enumerate • Low-Rank Approximation GG+G • Binary Features BG+G • Random Walk CG+G Rank • Co-Clustering M(GM T +G)+G • Clustered Matrix Decomp. (MG+G)(GM T +G)+G Select • Binary Matrix Factorization (BG+G)(GB T +G)+G • Dependent GSM (exp(GG+G)◦G)+G • ……

  24. Layer-wise Exhaustive Search GG + G M(exp(G) ∘ G) + G Enumerate MG + G G M(GG+G) + G Rank M(G T M+G) + G Select …… …… −1.32 +2.01 −1.32 +2.01 −2.33 0 0 0 0 1 0 0 0 0 1 −1.01 1.36 −1.64 +0.04 −0.23 +0.04 −0.23 −0.30 0 0 0 1 0 1 0 −0.22 0 0 0 1 0 −0.76 0.48 −0.50 + 12345 +0.08 +0.95 +0.08 +0.95 −1.27 + 12345 ⋮ ⋮ ⋮ ⋮ 0 1 −1.30 ⋮ −0.68 +0.52 −0.68 +0.52 −0.54 −0.24 0.89 −1.12 0 0 1 0 0 0 0 1 0 0 −0.77 +1.40 −0.77 +1.40 −1.66 Cluster Center Cluster Label Cluster Center Input Matrix 5×2 @(2×3) 20×3 20×5 5× 3 Structure ! Structure 6! + ! Structure 6(!! + !) + !

  25. Layer-wise Exhaustive Search Key Observation : Each step involves the same sub-problem. GG + G M(exp(G) ∘ G) + G Enumerate MG + G G M(GG+G) + G Rank M(G T M+G) + G Select …… …… −1.32 +2.01 −1.32 +2.01 −2.33 0 0 0 0 1 0 0 0 0 1 −1.01 1.36 −1.64 +0.04 −0.23 +0.04 −0.23 −0.30 0 0 0 1 0 1 0 −0.22 0 0 0 1 0 −0.76 0.48 −0.50 + 12345 +0.08 +0.95 +0.08 +0.95 −1.27 + 12345 ⋮ ⋮ ⋮ ⋮ 0 1 −1.30 ⋮ −0.68 +0.52 −0.68 +0.52 −0.54 −0.24 0.89 −1.12 0 0 1 0 0 0 0 1 0 0 −0.77 +1.40 −0.77 +1.40 −1.66 Cluster Center Cluster Label Cluster Center Input Matrix 5×2 @(2×3) 20×3 20×5 5× 3 Structure ! Structure 6! + ! Structure 6(!! + !) + !

Download Presentation
Download Policy: The content available on the website is offered to you 'AS IS' for your personal information and use only. It cannot be commercialized, licensed, or distributed on other websites without prior consent from the author. To download a presentation, simply click this link. If you encounter any difficulties during the download process, it's possible that the publisher has removed the file from their server.

Recommend

Library Learning for Neurally-Guided Bayesian Program Induction Kevin Ellis 1 , Lucas Morales 1 ,

Library Learning for Neurally-Guided Bayesian Program Induction Kevin Ellis 1 , Lucas Morales 1 ,

Library Learning for Neurally-Guided Bayesian Program Induction Kevin Ellis 1 , Lucas Morales 1 , Mathias Sabl e-Meyer 2 , Armando Solar-Lezama 1 , Joshua B. Tenenbaum 1 1 : MIT. 2 : ENS Paris-Saclay. Symbolic regression: List processing: Text

579 views • 21 slides
Knowledge Guided Attention and Inference for Describing Images Containing Unseen Objects Aditya

Knowledge Guided Attention and Inference for Describing Images Containing Unseen Objects Aditya

Knowledge Guided Attention and Inference for Describing Images Containing Unseen Objects Aditya Mogadala, Umanga Bista, Lexing Xie, Achim Rettinger rettinger@kit.edu, http://www.aifb.kit.edu/web/Achim_Rettinger/en,

520 views • 27 slides
Structure-Guided Discovery of ( S) -3 - Structure-Guided Discovery of ( S) -3 - ( am inom ethyl)

Structure-Guided Discovery of ( S) -3 - Structure-Guided Discovery of ( S) -3 - ( am inom ethyl)

OH B O ABX NH 2 Structure-Guided Discovery of ( S) -3 - Structure-Guided Discovery of ( S) -3 - ( am inom ethyl) benzo[ c] [ 1 ,2 ] oxaborol- ( am inom ethyl) benzo[ c] [ 1 ,2 ] oxaborol- 1 ( 3 H) -ol hydrochloride ( ABX) : A First in 1 ( 3

281 views • 25 slides
Guided Pathways Equity & Education Update Feb 7, 2020 Guided Pathways Decision Making

Guided Pathways Equity & Education Update Feb 7, 2020 Guided Pathways Decision Making

Guided Pathways Equity & Education Update Feb 7, 2020 Guided Pathways Decision Making Structure Equity & Education Council Guided Pathways Study Group Meta Majors Onboarding Communications ? Team Team Team Program Mapping,

289 views • 8 slides
Inference Suppose you are given a Bayesian network with the graph structure and the parameters

Inference Suppose you are given a Bayesian network with the graph structure and the parameters

Inference Suppose you are given a Bayesian network with the graph structure and the parameters CS 331: Artificial Intelligence all figured out Bayesian Networks (Inference) Now you would like to use it to do inference You need

417 views • 5 slides
Inference Suppose you are given a Bayesian network with the graph structure and the parameters

Inference Suppose you are given a Bayesian network with the graph structure and the parameters

CS 331: Artificial Intelligence Bayesian Networks (Inference) 1 Inference Suppose you are given a Bayesian network with the graph structure and the parameters all figured out Now you would like to use it to do inference You need

465 views • 13 slides
Reach and grasp by people with tetraplegia using a neurally controlled robotic arm Nature, 17

Reach and grasp by people with tetraplegia using a neurally controlled robotic arm Nature, 17

Leigh R. Hochberg et al. Reach and grasp by people with tetraplegia using a neurally controlled robotic arm Nature, 17 May 2012 Paper overview Ilya Kuzovkin 11 April 2014, Tartu etc How it works? etc Reach and grasp by people with

1.08k views • 51 slides
Pacemaker therapy for patients with neurally-mediated syncope and documented asystole A

Pacemaker therapy for patients with neurally-mediated syncope and documented asystole A

ISSUE 3 SYNCOPE ISSUE 3 I nternational S tudy on S yncope of U ncertain E tiology 3 Pacemaker therapy for patients with neurally-mediated syncope and documented asystole A randomized controlled double-blind trial ISSUE 3 Study design

139 views • 9 slides
AIRCRAFT WING STRUCTURE MONITORING USING AN INTEGRATED IMPEDANCE AND GUIDED WAVE TECHNIQUE Y.K. An

AIRCRAFT WING STRUCTURE MONITORING USING AN INTEGRATED IMPEDANCE AND GUIDED WAVE TECHNIQUE Y.K. An

18 TH INTERNATIONAL CONFERENCE ON COMPOSITE MATERIALS AIRCRAFT WING STRUCTURE MONITORING USING AN INTEGRATED IMPEDANCE AND GUIDED WAVE TECHNIQUE Y.K. An 1 *, H. Sohn 1 , C.Y. Park 2 1 Department of Civil and Environmental Engineering, KAIST,

406 views • 4 slides
FFR Guided Functional FFR Guided Functional FFR Guided Functional FFR Guided Functional

FFR Guided Functional FFR Guided Functional FFR Guided Functional FFR Guided Functional

FFR Guided Functional FFR Guided Functional FFR Guided Functional FFR Guided Functional Angioplasty Angioplasty g g p p y y in Complex Anatomy in Complex Anatomy Jung-Min Ahn, MD g , Heart Institute Asan Medical Center Heart

671 views • 27 slides
MVC Guided Pathways Brief review of Guided Pathways at MVC Plan for Today Spring

MVC Guided Pathways Brief review of Guided Pathways at MVC Plan for Today Spring

MVC Guided Pathways Brief review of Guided Pathways at MVC Plan for Today Spring 2018-Summer 2019 Action Plan and Budget Guided Pathways in California A Framework for Improving Completion Multiple Guided Pathways Initiatives

407 views • 13 slides
Coverage-Guided Fuzzing Dynamic Static Smart Coverage Structure Algorithms Security

Coverage-Guided Fuzzing Dynamic Static Smart Coverage Structure Algorithms Security

Coverage-Guided Fuzzing Dynamic Static Smart Coverage Structure Algorithms Security Testing Andreas Zeller, Saarland University Our Goal We want to cause the program to fail We have seen random (unstructured) input

868 views • 69 slides
Year 3 Guided Pathways Plan Presentation Presented by: Palomar Guided Pathways Team DATE: May

Year 3 Guided Pathways Plan Presentation Presented by: Palomar Guided Pathways Team DATE: May

Year 3 Guided Pathways Plan Presentation Presented by: Palomar Guided Pathways Team DATE: May 15, 2020 Presenters: Glyn Bongolan, Alex Cuatok, Kelly Falcone, Katy Farrell, Wendy Nelson And our Guided Pathways Ambassadors Focus for Today:

656 views • 50 slides
Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David

Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David

To appear, Proc. 15th International Joint Conference on AI (IJCAI-97), Nagoya, Japan, August, 1997 Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David Poole Department of Computer Science University

220 views • 8 slides
Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David

Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David

Probabilistic Partial Evaluation: Exploiting rule structure in probabilistic inference David Poole University of British Columbia 1 Overview Belief Networks Variable Elimination Algorithm Parent Contexts & Structured

259 views • 21 slides
Pattern matching and common structure inference in RNA (secondary) structures St ephane

Pattern matching and common structure inference in RNA (secondary) structures St ephane

Pattern matching and common structure inference in RNA (secondary) structures St ephane Vialette Stephane.Vialette@lri.fr Laboratoire de Recherche en Informatique (LRI) b at.490, Univ. Paris-Sud XI, 91405 Orsay cedex, France

794 views • 44 slides
Exhaustive search of optimal formulae for bilinear maps Svyatoslav Covanov Supervisors:

Exhaustive search of optimal formulae for bilinear maps Svyatoslav Covanov Supervisors:

Exhaustive search of optimal formulae for bilinear maps Svyatoslav Covanov Supervisors: Jrmie Detrey and Emmanuel Thom Team CARAMBA January 16, 2017 Optimal formulae 1 Previous work 2 Contribution and experimental results 3

390 views • 22 slides
Parsimony II Search Algorithms Genome 373 Genomic Informatics Elhanan Borenstein A quick

Parsimony II Search Algorithms Genome 373 Genomic Informatics Elhanan Borenstein A quick

Parsimony II Search Algorithms Genome 373 Genomic Informatics Elhanan Borenstein A quick review The parsimony principle: Find the tree that requires the fewest evolutionary changes! A fundamentally different method: Search

590 views • 18 slides
PROBABILISTIC ANALYSIS OF AN EXHAUSTIVE SEARCH ALGORITHM IN RANDOM GRAPHS Hsien-Kuei Hwang

PROBABILISTIC ANALYSIS OF AN EXHAUSTIVE SEARCH ALGORITHM IN RANDOM GRAPHS Hsien-Kuei Hwang

PROBABILISTIC ANALYSIS OF AN EXHAUSTIVE SEARCH ALGORITHM IN RANDOM GRAPHS Hsien-Kuei Hwang Academia Sinica, Taiwan (joint work with Cyril Banderier, Vlady Ravelomanana, Vytas Zacharovas) AofA 2008, Maresias, Brazil April 14, 2008 Hsien-Kuei

771 views • 59 slides
Introduction The MP Problem Solving a system of m multivariate polynomial equations in n variables

Introduction The MP Problem Solving a system of m multivariate polynomial equations in n variables

Fast Exhaustive Search for Quadratic Systems in F 2 on FPGAs Charles Bouillaguet, Chen-Mou Cheng, Tung Chou, Ruben Niederhagen, Bo- Yin Yang August 15, 2013 Introduction The MP Problem Solving a system of m multivariate polynomial equations in

790 views • 32 slides
Review Models of Interactivity mousePressed() mouseReleased() mouseClicked()

Review Models of Interactivity mousePressed() mouseReleased() mouseClicked()

Review Models of Interactivity mousePressed() mouseReleased() mouseClicked() OOP mouseMoved() encapsulation mouseDragged() inheritance keyPressed() keyReleased() keyTyped() mouseButton

247 views • 23 slides
15-251 Great Theoretical Ideas in Computer Science Lecture 17: NP and NP-completeness 1 March

15-251 Great Theoretical Ideas in Computer Science Lecture 17: NP and NP-completeness 1 March

15-251 Great Theoretical Ideas in Computer Science Lecture 17: NP and NP-completeness 1 March 21st, 2017 The big chasm between poly-time and exp-time. testing primality subset-sum matrix scheduling multiplication TSP MST Hamiltonian

780 views • 60 slides
10/16/19 Parameter Control Genetic Algorithms Motivation Parameter setting Tuning

10/16/19 Parameter Control Genetic Algorithms Motivation Parameter setting Tuning

10/16/19 Parameter Control Genetic Algorithms Motivation Parameter setting Tuning Control Examples Where to apply parameter control How to apply parameter control Parameter Control Motivation Motivation An EA

659 views • 7 slides
What is a predicate? A predicate is a statement involving variables over a specified

What is a predicate? A predicate is a statement involving variables over a specified

What is a predicate? A predicate is a statement involving variables over a specified domain (set). Example Domain (set) Predicate Integers ( Z ) S ( x ): x is a perfect square Reals ( R ) G ( x , y ): x > y Computers A ( c ): c

290 views • 11 slides

More recommend