search and machine learning
play

Search and Machine Learning Kalyan Veeramachaneni, Jason Ansel, - PowerPoint PPT Presentation

Nov 24, 2022 •21 likes •371 views

Search and Machine Learning Kalyan Veeramachaneni, Jason Ansel, Shoaib Kamil, Jeffrey Bosboom, Una-May OReilly, Saman Amarasinghe CGO Tutorial February 8th, 2015 Overview Parameter types and tuning An example tuning problem -

  1. � Search and Machine Learning Kalyan Veeramachaneni, Jason Ansel, Shoaib Kamil, Jeffrey Bosboom, Una-May O’Reilly, Saman Amarasinghe CGO Tutorial February 8th, 2015

  2. Overview • Parameter types and tuning • An example tuning problem - permutation • A typical population based search process • Select-Create-Update process • Options • Create —> Operators • Select - Update • Composition of multiple search approaches • Steps to take for a new problem • Back to Mario example

  3. Parameter types

  4. Parameter types Let’s pick permutation Example: Permute the placement of these computations

  5. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 2 3 4 5 6 8 9 7 4 2 5 8 9 6 7 3 1 6 2 5 9 8 7 4 1 3 . . . 1 4 8 2 5 3 9 7 6 8 5 7 1 2 4 9 3 6 . . . 6 2 5 3 4 8 7 1 9

  6. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 2 3 4 5 6 7 8 9 0.97 4 2 5 8 0.52 9 6 7 3 1 9 8 7 4 0.32 1 6 2 5 3 . . . 2 1 5 3 9 7 4 6 8 0.73 8 5 7 1 2 4 9 3 6 0.84 . . . 6 2 5 3 7 4 1 8 9 0.14

  7. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 2 3 4 5 6 7 8 9 0.97 4 2 5 8 0.52 9 6 7 3 1 6 2 5 9 8 7 4 0.32 1 3 . . 1 2 3 4 5 6 7 8 9 . 8 5 7 1 2 4 9 3 6 2 1 5 3 9 7 4 6 8 0.73 8 5 7 1 2 4 9 3 6 0.84 . . . 6 2 5 3 7 4 1 8 9 0.14

  8. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 2 3 4 5 6 8 9 0.97 7 9 6 4 7 3 2 5 1 8 0.52 9 8 4 0.32 1 6 2 5 3 7 . . 1 2 3 4 5 6 7 8 9 9 6 3 4 5 8 7 1 2 . 8 5 7 1 2 4 9 3 6 6 8 7 1 2 9 3 4 5 2 1 5 3 9 7 4 6 8 0.73 8 5 7 1 2 4 9 3 6 0.84 . . . 6 2 5 3 4 1 8 9 0.14 7

  9. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 9 0.97 9 6 3 4 5 8 7 1 2 2 3 4 5 6 7 8 6 8 7 1 2 9 3 4 5 9 6 4 7 3 2 5 1 8 0.52 9 8 4 0.32 1 6 2 5 3 7 5 6 8 3 9 1 7 2 4 4 2 5 8 . 9 6 7 3 1 . . 1 4 8 7 8 6 9 5 2 1 4 3 2 5 3 9 7 6 2 1 5 3 9 7 4 6 8 0.73 8 5 7 1 2 4 9 3 6 0.84 . . . 6 2 5 3 4 1 8 0.14 7 9

  10. A typical population based search process Initialize Evaluate Select Create Evaluate Update 1 2 3 4 5 6 7 8 9 0.97 1 9 6 3 4 5 8 7 1 2 9 2 3 4 5 6 7 8 4 2 5 8 0.52 9 6 7 3 1 6 8 7 4 1 2 2 9 3 5 4 5 8 9 6 7 3 1 9 8 7 4 0.32 1 6 2 5 3 5 6 1 8 3 9 7 2 4 . . . 7 8 6 9 5 2 1 4 3 . . 1 4 8 2 5 3 9 7 6 0.73 . 8 5 7 1 2 4 9 3 6 0.84 . . . . . . 6 2 7 3 4 9 5 8 1 6 2 5 3 4 8 0.14 7 1 9

  11. A typical population based search process Select Create Initialize Evaluate Evaluate Update 1 2 3 4 5 6 8 9 0.97 7 9 6 3 4 5 8 7 1 2 1 2 3 4 5 6 7 8 9 0.94 4 2 5 1 8 0.52 9 6 7 3 9 6 6 8 7 4 7 1 3 2 2 9 3 5 1 4 8 5 0.87 9 8 4 0.32 6 2 5 7 6 8 3 9 0.73 1 3 5 1 7 2 4 . . . 7 8 6 9 5 2 1 4 3 0.53 . . 1 4 8 2 5 3 9 7 6 0.73 . 8 5 7 1 2 4 9 3 6 0.84 . . . . . . 6 2 7 3 4 9 5 8 1 0.12 3 4 0.14 6 2 5 7 1 8 9

  12. A typical population based search process Select Create Initialize Evaluate Evaluate Update 1 9 0.97 1 2 3 4 5 6 7 8 9 0.97 2 3 4 5 6 7 8 1 9 6 3 4 5 8 7 1 2 2 3 4 5 6 8 9 0.94 7 4 2 5 8 0.52 9 6 7 3 1 6 8 7 4 1 2 2 9 3 5 4 8 5 0.87 6 8 7 4 1 2 2 9 5 3 4 8 5 0.87 9 6 7 3 1 9 6 7 3 1 9 8 7 4 0.32 1 6 2 5 3 5 6 1 8 3 9 7 2 4 0.73 5 6 1 8 3 9 7 2 4 0.73 . . . 7 8 6 9 5 2 1 4 3 0.53 . . . . . 1 4 8 2 5 3 9 7 6 0.73 . . . . 8 5 7 1 2 4 9 3 6 0.84 . . . . . . 6 2 7 3 4 9 5 8 1 0.12 4 4 6 2 5 3 7 1 8 9 0.14 6 2 5 3 7 1 8 9 0.14

  13. A typical population based search process Select Create Initialize Evaluate Evaluate Update Continue

  14. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6

  15. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 _ _ │ 3 4 5 │ _ _ _ _ Copy over a portion of good material from #1

  16. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 Copy over a portion of good material from #1 _ _ │ 3 4 5 │ _ _ _ _

  17. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 9 Copy over a portion of good material from #1 _ _ │ 3 4 5 │ _ _ _ _

  18. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 9 6 Copy over a portion of good material from #1 _ _ │ 3 4 5 │ _ _ _ _

  19. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 9 6 8 Copy over a portion of good material from #1 _ _ │ 3 4 5 │ _ _ _ _

  20. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 9 6 8 7 1 2 Copy over a portion of good material from #1 _ _ │ 3 4 5 │ _ _ _ _

  21. Create —> Requires operators • Operators take two or more parameters and create two or more new parameters • Let’s take the permutation example: 1 2 3 4 5 6 7 8 9 � Input 8 5 7 1 2 4 9 3 6 1 2 │ 3 4 5 │ 6 7 8 9 cross over points 8 5 │ 7 1 2 │ 4 9 3 6 9 6 8 7 1 2 9 6 │ 3 4 5 │ 8 7 1 2 Copy over a portion of good material from #1

  22. Create: 
 Several options exist for operators • For permutation parameter there are several operators • Partially mapped crossover • Partition crossover • Ordered crossover • Edge crossover • Cycle crossover

  23. Select-update • Select —> chooses the solutions among the population from which new solutions will be created • biasing the search towards to better solutions • Update—> updates the entire population towards better search spaces • has stronger influence on convergence

  24. Particle Swarm Optimization (PSO) Select-update • State based—> for each member of the population, a history is maintained • Select —> Individual based • for every individual select its previous best • select the best solution seen so far • Update—> only update individuals history if it finds a better solution in the search space

  25. Genetic Algorithms Select-update • Select —> Tournament selection • select randomly k from n • among these k select the top 2 • allows enough mixing random k n best 2 k=5 1 2 3 4 5 6 8 9 0.97 7 9 6 4 7 3 2 5 1 8 0.52 9 8 4 0.32 1 6 2 5 3 7 . . . 1 4 8 2 5 3 9 7 6 0.73 8 5 7 1 2 4 9 3 6 0.84 . . . 6 2 5 3 4 8 0.14 7 1 9

  26. Genetic Algorithms Select-update • Update—> multiple ways that allow us to control exploration and exploitation • strong elitism • combine both old and new and select the top n • weak elitism • do it on a per individual basis, select if the new one it created is better than itself.

  27. Composition of multiple approaches GA - Genetic algorithms PSO- Particle swarm optimization For permutation DE- Differential evolution Operators Select-update choices Ordered PSO select-update Partition GA Partial select-update Cycle DE select-update Edge

  28. Composition of multiple approaches GA - Genetic algorithms PSO- Particle swarm optimization For permutation DE- Differential evolution Operators Select-update choices Ordered Ordered —PSO PSO select-update Partition GA Partial select-update Cycle DE select-update Edge

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

Example 1 : Selecting a present for somebody You enter a large shop with no idea what

Example 1 : Selecting a present for somebody You enter a large shop with no idea what

Evolutionary Computing (Genetic Algorithms) Heralded as an approach to machine learning Learning new solutions/behaviours by evolving old ones Also an approach to search Search in a potentially infinite search space No

321 views • 6 slides
Machine learning techniques to probe theoretical physics Intro In inSPIRE, search find t

Machine learning techniques to probe theoretical physics Intro In inSPIRE, search find t

Akinori Tanaka (RIKEN AIP/iTHEMS) Machine learning techniques to probe theoretical physics Intro In inSPIRE, search find t machine learning OR deep learning and date 20xx->20xx+1 year number of results Intro Mainly experimental, a few

908 views • 70 slides
Leveraging AWS and Machine Learning to Power Search at Zocdoc Pedro Rubio Head of Search

Leveraging AWS and Machine Learning to Power Search at Zocdoc Pedro Rubio Head of Search

Leveraging AWS and Machine Learning to Power Search at Zocdoc Pedro Rubio Head of Search Engineering Brian dAlessandro Head of Data Science This document and its contents are proprietary and confidential of Zocdoc, Inc. and may not be

582 views • 35 slides
Learning to Search + Recurrent Neural Networks Matt Gormley Lecture 4 Sep. 9, 2019 1

Learning to Search + Recurrent Neural Networks Matt Gormley Lecture 4 Sep. 9, 2019 1

10-418 / 10-618 Machine Learning for Structured Data Machine Learning Department School of Computer Science Carnegie Mellon University Learning to Search + Recurrent Neural Networks Matt Gormley Lecture 4 Sep. 9, 2019 1 Reminders

599 views • 58 slides
Hyperparameter Search in Machine Learning Marc Claesen and Bart De Moor

Hyperparameter Search in Machine Learning Marc Claesen and Bart De Moor

Introduction Example: optimizing hyperparameters for an SVM classifier Challenges in hyperparameter search State-of-the-art References Hyperparameter Search in Machine Learning Marc Claesen and Bart De Moor marc.claesen@esat.kuleuven.be

770 views • 57 slides
An Exercise in An Exercise in Machine Learning Machine Learning

An Exercise in An Exercise in Machine Learning Machine Learning

An Exercise in An Exercise in Machine Learning Machine Learning http://www.cs.iastate.edu/~cs573x/bbsilab.html Machine Learning Software Preparing Data Building Classifiers Interpreting Results Machine Learning Software

496 views • 26 slides
Machine Learning By Alex Scarlatos What is Machine Learning? Machine Learning is the process by

Machine Learning By Alex Scarlatos What is Machine Learning? Machine Learning is the process by

Machine Learning By Alex Scarlatos What is Machine Learning? Machine Learning is the process by which computers can be trained through observation, rather than being explicitly programmed. Basically, a program is given input and adjusts its

556 views • 17 slides
Machine Learning: Study of algorithms that improve their performance P at some task T

Machine Learning: Study of algorithms that improve their performance P at some task T

Machine Learning 10-601 Tom M. Mitchell Machine Learning Department Carnegie Mellon University January 12, 2015 Today: Readings: What is machine learning? The Discipline of ML Decision tree learning Mitchell, Chapter 3

872 views • 29 slides
Traditional Machine Learning: Unsupervised Learning Juhan Nam Traditional Machine Learning

Traditional Machine Learning: Unsupervised Learning Juhan Nam Traditional Machine Learning

GCT634/AI613: Musical Applications of Machine Learning (Fall 2020) Traditional Machine Learning: Unsupervised Learning Juhan Nam Traditional Machine Learning Pipeline in Classification Tasks A set of hand-designed audio features are selected

398 views • 26 slides
10d Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10d Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10d Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A Framework for 10.6 Unsupervised Learning Symbol-based Learning 10.7 Reinforcement Learning 10.2 Version Space Search 10.8 Epilogue and 10.3 The

1.05k views • 29 slides
10b Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10b Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10b Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A Framework for 10.6 Unsupervised Learning Symbol-based Learning 10.7 Reinforcement Learning 10.2 Version Space Search 10.8 Epilogue and 10.3 The

866 views • 50 slides
10c Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10c Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10c Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A Framework for 10.6 Unsupervised Learning Symbol-based Learning 10.7 Reinforcement Learning 10.2 Version Space Search 10.8 Epilogue and 10.3 The

605 views • 35 slides
10a Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10a Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A

10a Machine Learning: Symbol-based 10.0 Introduction 10.5 Knowledge and Learning 10.1 A Framework for 10.6 Unsupervised Learning Symbol-based Learning 10.7 Reinforcement Learning 10.2 Version Space Search 10.8 Epilogue and 10.3 The

934 views • 54 slides
Retrieval as Interaction African Summer School on Machine Learning for Data Mining and Search

Retrieval as Interaction African Summer School on Machine Learning for Data Mining and Search

Retrieval as Interaction African Summer School on Machine Learning for Data Mining and Search Maarten de Rijke January 14, 2019 University of Amsterdam derijke@uva.nl Based on joint work with Abhinav Khaitan, Ana Lucic, Anne Schuth, Boris

1.16k views • 50 slides
CS 335 Machine Learning What is Machine Learning? Dan Sheldon Spring 2019 What is Machine

CS 335 Machine Learning What is Machine Learning? Dan Sheldon Spring 2019 What is Machine

1/22/19 CS 335 Machine Learning What is Machine Learning? Dan Sheldon Spring 2019 What is Machine Learning? A Simple Task: Recognize Obama How do you program a computer to Recognize faces? Recommend movies? Decide which web

581 views • 5 slides
Machine Learning Machine Learning: algorithms that use experience to improve their

Machine Learning Machine Learning: algorithms that use experience to improve their

Machine Learning Machine Learning: algorithms that use experience to improve their performance We use machine learning in situations where it is very challenging (or impossible) to define the rules by hand: e.g. face detection

1.04k views • 28 slides
Forecasting Non-Stationary Time Series without Recurrent Connections AP Engelbrecht Department

Forecasting Non-Stationary Time Series without Recurrent Connections AP Engelbrecht Department

Forecasting Non-Stationary Time Series without Recurrent Connections AP Engelbrecht Department of Industrial Enigneering, and Computer Science Division Stellenbosch University South Africa engel@sun.ac.za Engelbrecht (Stellenbosch

484 views • 30 slides
Aerial Swarm Robotics Introduction and Exploration Parth Sarthi Pandey Universitt Hamburg

Aerial Swarm Robotics Introduction and Exploration Parth Sarthi Pandey Universitt Hamburg

MIN-Fakultt Fachbereich Informatik Aerial Swarm Robotics Introduction and Exploration Parth Sarthi Pandey Universitt Hamburg Fakultt fr Mathematik, Informatik und Naturwissenschaften Fachbereich Informatik Technische Aspekte

761 views • 25 slides
Strategic planning in ATM with a stochastic anytime approach J. A. Cobano, D. Alejo, G. Heredia

Strategic planning in ATM with a stochastic anytime approach J. A. Cobano, D. Alejo, G. Heredia

Strategic planning in ATM with a stochastic anytime approach J. A. Cobano, D. Alejo, G. Heredia and A. Ollero Robotics, Vision and Control Group, University of Seville (Spain) Strategic planning in ATM with a stochastic anytime approach

487 views • 31 slides
Optimization of Polytopic System Eigenvalues by Swarm of Particles Jacek Kabzi ski, Jaros aw

Optimization of Polytopic System Eigenvalues by Swarm of Particles Jacek Kabzi ski, Jaros aw

Optimization of Polytopic System Eigenvalues by Swarm of Particles Jacek Kabzi ski, Jaros aw Kacerka Institute of Automatic Control Lodz Univeristy of Technology, Poland Plan Introduction Eigenvalue optimization Linear

309 views • 18 slides
Emergent Optimization: Design and Applications in Telecommunications and Bioinformatics PhD

Emergent Optimization: Design and Applications in Telecommunications and Bioinformatics PhD

Emergent Optimization: Design and Applications in Telecommunications and Bioinformatics PhD Thesis Dissertation Author: Jos Manuel Garca-Nieto Advisor: Dr. Enrique Alba PhD Thesis Dissertation Jos Manuel Garca-Nieto 1 / 53

555 views • 53 slides
Particle Swarm Optimization for Voltage Stability Analysis Dinesh Rangana Gurusinghe University

Particle Swarm Optimization for Voltage Stability Analysis Dinesh Rangana Gurusinghe University

Electrical Power and Energy Conference 2012 Resilient Green Energy Systems for a Sustainable Society Particle Swarm Optimization for Voltage Stability Analysis Dinesh Rangana Gurusinghe University of Manitoba, Canada Weerakorn Ongsakul Asian

581 views • 28 slides
Mrs: High Performance MapReduce for Iterative and Asynchronous Algorithms in Python Jeff Lund ,

Mrs: High Performance MapReduce for Iterative and Asynchronous Algorithms in Python Jeff Lund ,

Mrs. Iterative MapReduce Performance and Case Studies Mrs: High Performance MapReduce for Iterative and Asynchronous Algorithms in Python Jeff Lund , Chace Ashcraft, Andrew McNabb and Kevin Seppi Brigham Young University November 14, 2016

589 views • 23 slides
L101: Optimization fundamentals Previous lecture Logistic regression parameter learning:

L101: Optimization fundamentals Previous lecture Logistic regression parameter learning:

L101: Optimization fundamentals Previous lecture Logistic regression parameter learning: Supervised machine learning algorithms typically involve optimizing a loss over the training data: This is an instance of numerical optimization , i.e.

486 views • 21 slides

More recommend