reinforcement learning
play

Reinforcement Learning a gentle introduction & industrial - PowerPoint PPT Presentation

Jan 08, 2023 •28 likes •618 views

Reinforcement Learning a gentle introduction & industrial application Christian Hidber Learning learning from children FOLIE 2 REINFORCEMENT LEARNING The game: demo FOLIE 3 REINFORCEMENT LEARNING The game: setup game engine Goal:

  1. Reinforcement Learning a gentle introduction & industrial application Christian Hidber

  2. Learning learning from children FOLIE 2 REINFORCEMENT LEARNING

  3. The game: demo FOLIE 3 REINFORCEMENT LEARNING

  4. The game: setup game engine Goal: maximize sum of rewards Step reward actions game state learner FOLIE 4 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  5. The game: positive feedback game engine Step reward actions game state learner FOLIE 5 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  6. The game: negative feedback game engine Step reward actions game state learner FOLIE 6 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  7. the learned stuff => policy game engine Step reward actions game state Policy learner (rules learned, how to play the game) FOLIE 7 REINFORCEMENT LEARNING

  8. policy improvement => learning game engine Step reward actions game state Policy learner (rules learned, how to play the game) FOLIE 8 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  9. policy improvement => learning game engine Step reward actions game state Policy learner (rules learned, how to play the game) FOLIE 9 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  10. Reinforcement learning game engine Key idea: continuously improve policy to increase total reward Step reward actions game state Policy RL algorithm (rules learned, how to play the game) FOLIE 10 REINFORCEMENT LEARNING "Dieses Foto" von Unbekannter Autor ist lizenziert gemäß CC BY-NC

  11. Episode 1 : play with 1st policy (random) Policy (rules learned, how to play the game) Action from State Policy Next State Reward -1 1 2 3 4 5 6 7 Step # FOLIE 11 REINFORCEMENT LEARNING

  12. Episode 1 : play with 1st policy (random) Policy (rules learned, how to play the game) Action from State Policy Next State Reward 100 1 2 3 4 5 6 7 Step # FOLIE 12 REINFORCEMENT LEARNING

  13. Episode 1 : play with 1st policy (random) Policy (rules learned, how to play the game) Action from State Policy Next State Reward -50 Episode Over Episode Over 1 2 3 4 5 6 7 Step # FOLIE 13 REINFORCEMENT LEARNING

  14. Episode 1 : improve 1st policy for state in step 3 Policy (rules learned, how to play the game) Action from State Policy -50 Episode Over 1 2 3 4 5 6 7 Step # FOLIE 14 REINFORCEMENT LEARNING

  15. Episode 1 : improve 1st policy for state in step 2 Policy (rules learned, how to play the game) Action from State Policy 50 (=+100 -50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 15 REINFORCEMENT LEARNING

  16. Episode 1 : improve 1st policy for state in step 1 Policy (rules learned, how to play the game) Action from State Policy 49 (=-1 +100 -50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 16 REINFORCEMENT LEARNING M3, OCTOBER 2018

  17. Episode 2 : play with 2nd policy Policy (rules learned, how to play the game) Already learned: go left is ok Action from State Policy Next State Reward -1 1 2 3 4 5 6 7 Step # FOLIE 17 REINFORCEMENT LEARNING

  18. Episode 2 : play with 2nd policy Policy (rules learned, how to play the game) Already learned: go left is ok Action from State Policy Next State Reward 100 1 2 3 4 5 6 7 Step # FOLIE 18 REINFORCEMENT LEARNING

  19. Already learned: Episode 2 : play with 2nd policy Policy don’t go up (rules learned, how to play the game) Action from State Policy Next State Reward 100 1 2 3 4 5 6 7 Step # FOLIE 19 REINFORCEMENT LEARNING

  20. Episode 2 : play with 2nd policy Policy (rules learned, how to play the game) Action from State Policy Next State Reward -1 1 2 3 4 5 6 7 Step # FOLIE 20 REINFORCEMENT LEARNING

  21. Episode 2 : play with 2nd policy Policy (rules learned, how to play the game) Action from State Policy Next State Reward -50 Episode Over Episode Over 1 2 3 4 5 6 7 Step # FOLIE 21 REINFORCEMENT LEARNING

  22. Episode 2 : improve 2nd policy for state in step 5 Policy (rules learned, how to play the game) Action from State Policy -50 (=-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 22 REINFORCEMENT LEARNING

  23. Episode 2 : improve 2nd policy for state in step 4 Policy (rules learned, how to play the game) Action from State Policy -51 (=-1-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 23 REINFORCEMENT LEARNING

  24. Episode 2 : improve 2nd policy for state in step 3 Policy (rules learned, how to play the game) Action from State Policy 49 (=+100-1-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 24 REINFORCEMENT LEARNING

  25. Episode 2 : improve 2nd policy for state in step 2 Policy (rules learned, how to play the game) Action from State Policy 149 (=+100+100-1-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 25 REINFORCEMENT LEARNING

  26. Episode 2 : improve 2nd policy for state in step 2 Policy (rules learned, how to play the game) Action from State Policy = some running average of old and new value 149 (=+100+100-1-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 26 REINFORCEMENT LEARNING

  27. Episode 2 : improve 2nd policy for state in step 1 Policy (rules learned, how to play the game) Action from State Policy 148 (=-1+100+100-1-50) Future Reward (sum of all rewards from current state until ‘game over’) Episode Over 1 2 3 4 5 6 7 Step # FOLIE 27 REINFORCEMENT LEARNING

  28. So far ….. a policy is a map from states to action probabilities FOLIE 28 REINFORCEMENT LEARNING

  29. …updated by the reinforcement learning algorithm a policy is a map from states to action probabilities FOLIE 29 REINFORCEMENT LEARNING

  30. …updated by the reinforcement learning algorithm a policy is a map from states to action probabilities FOLIE 30 REINFORCEMENT LEARNING

  31. After many, many episodes, for each state… FOLIE 31 REINFORCEMENT LEARNING

  32. Algorithm sketch Policy (rules learned, how to play the game) Initialize table with random action probabilities for each state Repeat play episode with policy given by table Record (state 1 ,action 1 ,reward 1 ),(state 2 ,action 2 ,reward 2 ),…. for episode For each step i compute FutureReward i = reward i + reward i+1 +… update table[state i ] s.t. • action i becomes for state i more likely if FutureReward i is “high” • action i becomes for state i less likely if FutureReward i is “low” FOLIE 32 REINFORCEMENT LEARNING

  33. Algorithm sketch Policy (rules learned, how to play the game) Initialize table with random action probabilities for each state Repeat play episode with policy given by table Record (state 1 ,action 1 ,reward 1 ),(state 2 ,action 2 ,reward 2 ),…. for episode For each step i compute FutureReward i = reward i + reward i+1 +… update table[state i ] s.t. • action i becomes for state i more likely if FutureReward i is “high” • action i becomes for state i less likely if FutureReward i is “low” FOLIE 33 REINFORCEMENT LEARNING

  34. Algorithm sketch Policy (rules learned, how to play the game) Initialize table with random action probabilities for each state Repeat play episode with policy given by table Record (state 1 ,action 1 ,reward 1 ),(state 2 ,action 2 ,reward 2 ),…. for episode For each step i compute FutureReward i = reward i + reward i+1 +… update table[state i ] s.t. • action i becomes for state i more likely if FutureReward i is “high” • action i becomes for state i less likely if FutureReward i is “low” FOLIE 34 REINFORCEMENT LEARNING

  35. The game: demo FOLIE 35 REINFORCEMENT LEARNING

  36. The bad news: nice idea, but… FOLIE 36 «Image" licensed according to CC BY-SA REINFORCEMENT LEARNING

  37. The bad news: nice idea, but… too many states… too many actions • Too much memory needed • Too much time FOLIE 37 «Image" licensed according to CC BY-SA REINFORCEMENT LEARNING

  38. The solution Policy (rules learned, how to play the game) Idea: Replace lookup table with a neural network that approximates the action probabilities contained in the table Instead of Table[state] = action probabilities Do NeuralNet( state ) ~ action probablities Change to “play episode with policy given by NeuralNet” Change to “update weights of NeuralNet” FOLIE 38 REINFORCEMENT LEARNING

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

RL Overview of topics About Reinforcement Learning The Reinforcement Learning Problem

RL Overview of topics About Reinforcement Learning The Reinforcement Learning Problem

Introduction to Reinforcement Learning RL Overview of topics About Reinforcement Learning The Reinforcement Learning Problem Inside an RL agent Temporal difference learning Many faces of Reinforcement Learning What is

551 views • 35 slides
Reinforcement Learning Reinforcement Learning Reinforcement Learning in a nutshell g Imagine

Reinforcement Learning Reinforcement Learning Reinforcement Learning in a nutshell g Imagine

Reinforcement Learning Reinforcement Learning Reinforcement Learning in a nutshell g Imagine playing a new game whose rules you dont know; after a hundred or so moves your don t know; after a hundred or so moves, your opponent announces, You

511 views • 30 slides
Reinforcement Learning You can think of supervised learning as the teacher providing answers

Reinforcement Learning You can think of supervised learning as the teacher providing answers

Reinforcement Learning You can think of supervised learning as the teacher providing answers (the class labels) Reinforcement learning: In reinforcement learning, the agent learns Markov Decision Processes based on a punishment/reward

89 views • 6 slides
Examples of Reinforcement Learning Robocup Soccer Teams Stone & Veloso, Reidmiller et al.

Examples of Reinforcement Learning Robocup Soccer Teams Stone & Veloso, Reidmiller et al.

Reinforcement Learning Slides by Rich Sutton Mods by Dan Lizotte Refer to Reinforcement Learning: An Introduction by Sutton and Barto Alpaydin Chapter 16 Up until now we have been Supervised Learning Classifying, mostly

891 views • 31 slides
CS885 Reinforcement Learning Module 2: June 6, 2020 Maximum Entropy Reinforcement Learning

CS885 Reinforcement Learning Module 2: June 6, 2020 Maximum Entropy Reinforcement Learning

CS885 Reinforcement Learning Module 2: June 6, 2020 Maximum Entropy Reinforcement Learning Haarnoja, Tang et al. (2017) Reinforcement Learning with Deep Energy Based Policies, ICML . Haarnoja, Zhou et al. (2018) Soft Actor-Critic: Off-Policy

682 views • 24 slides
Reinforcement Learning Reinforcement Learning Now that you know a little about Optimal Control

Reinforcement Learning Reinforcement Learning Now that you know a little about Optimal Control

Reinforcement Learning Reinforcement Learning Now that you know a little about Optimal Control Theory, you actually have some knowledge in RL. RL shares the overall goal with OCT: solving for a control policy such that the cumulative cost

610 views • 22 slides
Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning II Still

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning II Still

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning II Still assume a Marko kov decision process (MDP): A se set o of st states s s S A se set o of a actions ( s (per st state) A A mo model

96 views • 5 slides
R i f R i f Reinforcement Learning III Reinforcement Learning III t L t L i i III III Dec

R i f R i f Reinforcement Learning III Reinforcement Learning III t L t L i i III III Dec

R i f R i f Reinforcement Learning III Reinforcement Learning III t L t L i i III III Dec 03 2008 1 Large State Spaces h When a problem has a large state space we can not longer represent the U or Q functions as explicit tables explicit

436 views • 18 slides
Reinforcement Learning UMaine COS 470/570 Introduction to AI Why reinforcement learning?

Reinforcement Learning UMaine COS 470/570 Introduction to AI Why reinforcement learning?

Reinforcement Learning<br/><br/> 4/25/19, 8*06 PM Reinforcement Learning<br/><br/> 4/25/19, 8*06 PM Reinforcement Learning UMaine COS 470/570 Introduction to AI Why reinforcement learning? Spring 2019 Created:

371 views • 15 slides
7. Motor Control and Reinforcement Learning Outline A. Action Selection and Reinforcement B.

7. Motor Control and Reinforcement Learning Outline A. Action Selection and Reinforcement B.

7. Motor Control and Reinforcement Learning Outline A. Action Selection and Reinforcement B. Temporal Difference Reinforcement Learning C. PVLV Model D. Cerebellum and Error-driven Learning 2/23/18 COSC 494/594 CCN 2 Sensory-Motor Loop

790 views • 56 slides
Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree

Deep Reinforcement Learning [Mastering the Game of Go with Deep Reinforcement Learning and Tree Search, Nature 2016] CS 486/686 University of Waterloo Lecture 21: July 12, 2017 Outline AlphaGo Supervised Learning of Policy Networks

540 views • 15 slides
Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning Ja Jan-Wi

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning Ja Jan-Wi

Reinforcement Learning CS 4100: Artificial Intelligence Reinforcement Learning Ja Jan-Wi Willem van de Meent Northeastern University [These slides were created by Dan Klein and Pieter Abbeel for CS188 Intro to AI at UC Berkeley. All CS188

145 views • 10 slides
Reinforcement Learning AIMA Chapters: 21.1, 21.2, 21.3. Sutton and Barto, Reinforcement Learning:

Reinforcement Learning AIMA Chapters: 21.1, 21.2, 21.3. Sutton and Barto, Reinforcement Learning:

Reinforcement Learning Reinforcement Learning AIMA Chapters: 21.1, 21.2, 21.3. Sutton and Barto, Reinforcement Learning: an Introduction, 2nd Edition: Chapters 6 (6.1 6.5) Outline Reinforcement Learning Reinforcement Learning: the

587 views • 27 slides
Module 11 Introduction to Reinforcement Learning CS 886 Sequential Decision Making and

Module 11 Introduction to Reinforcement Learning CS 886 Sequential Decision Making and

Module 11 Introduction to Reinforcement Learning CS 886 Sequential Decision Making and Reinforcement Learning University of Waterloo Machine Learning Supervised Learning Teacher tells learner what to remember Reinforcement Learning

401 views • 29 slides
1 Pure Reinforcement Learning vs. Reinforcement Learning Monte-Carlo Planning No knowledge

1 Pure Reinforcement Learning vs. Reinforcement Learning Monte-Carlo Planning No knowledge

CSE 573: Artificial Intelligence Todo Reinforcement Learning Add simulations from 473 Dan Weld Add UCB bound (cut bolzman & constant epsilon Add snazzy videos (pendulum, zico kolter See http://www See http://www-

462 views • 11 slides
Safe Reinforcement Learning Philip S. Thomas Stanford CS234: Reinforcement Learning, Guest

Safe Reinforcement Learning Philip S. Thomas Stanford CS234: Reinforcement Learning, Guest

Safe Reinforcement Learning Philip S. Thomas Stanford CS234: Reinforcement Learning, Guest Lecture May 24, 2017 Lecture overview What makes a reinforcement learning algorithm safe ? Notation Creating a safe reinforcement learning

1.42k views • 88 slides
Health Care Spending and Utilization: Evidence from Perinatal Care in Arkansas C A I T L I N C A

Health Care Spending and Utilization: Evidence from Perinatal Care in Arkansas C A I T L I N C A

Effects of Episode-Based Payment on Health Care Spending and Utilization: Evidence from Perinatal Care in Arkansas C A I T L I N C A R R O L L P H D C A N D I D A T E H A R V A R D U N I V E R S I T Y J O I N T W O R K W I T H M . C H E

251 views • 14 slides
Quality Improvement Techniques for Public Health: Useful in Addressing the Substance Use Disorder

Quality Improvement Techniques for Public Health: Useful in Addressing the Substance Use Disorder

Quality Improvement Techniques for Public Health: Useful in Addressing the Substance Use Disorder Crisis? Angela Carman, DrPH March 2020 What is Quality Improvement? Systematic problem solving required to develop and improve products

206 views • 18 slides
Integrated Disease Surveillance Programme Under umbrella of National Rural Health Mission

Integrated Disease Surveillance Programme Under umbrella of National Rural Health Mission

Integrated Disease Surveillance Programme Under umbrella of National Rural Health Mission (NRHM) Objective: Strengthening of Disease Surveillance System for epidemic prone diseases to detect and respond to outbreaks Integrated Disease

210 views • 7 slides
Public Data Analytics: The New Essential Service Francesco Mureddu Director - The Lisbon Council

Public Data Analytics: The New Essential Service Francesco Mureddu Director - The Lisbon Council

Public Data Analytics: The New Essential Service Francesco Mureddu Director - The Lisbon Council Brussels, 26 June 2020 Task 2 - Policy Modelling and Simulation Five simulation models: finance, energy, pollution and climate change. What is

552 views • 9 slides
Client Alert CMS Announces Bundled Payments for Care Improvement initiative Contact Attorney

Client Alert CMS Announces Bundled Payments for Care Improvement initiative Contact Attorney

Client Alert CMS Announces Bundled Payments for Care Improvement initiative Contact Attorney Regarding in Efgort to Save Money and Improve Care This Matter: R. Michael Barry On August 23, 2011, the Center for Medicare & Medicaid Services

412 views • 3 slides
2018 SALES EXCEED CHF 7 BILLION FOR THE FIRST TIME SIKA MADE BINDING OFFER TO ACQUIRE PAREX SIKA

2018 SALES EXCEED CHF 7 BILLION FOR THE FIRST TIME SIKA MADE BINDING OFFER TO ACQUIRE PAREX SIKA

2018 SALES EXCEED CHF 7 BILLION FOR THE FIRST TIME SIKA MADE BINDING OFFER TO ACQUIRE PAREX SIKA INVESTOR PRESENTATION JANUARY 2019 AGENDA 1. HIGHLIGHTS SALES 2018 2. SUCCESSFUL STRATEGY EXECUTION 3. SIKA MADE BINDING OFFER TO ACQUIRE

399 views • 39 slides
Improving Educational Outcomes and New Mexico Response to Yazzie and Martinez v State of New

Improving Educational Outcomes and New Mexico Response to Yazzie and Martinez v State of New

Improving Educational Outcomes and New Mexico Response to Yazzie and Martinez v State of New Mexico Decision Legislative Finance Committee Ryan Stewart, Secretary Public Education Department Aug. 28, 2020 Investing for tomorrow, delivering

467 views • 29 slides
Community Participation in Research STREAM Clinical Trial [INSERT NAME OF COUNTRY] 2 Why do we

Community Participation in Research STREAM Clinical Trial [INSERT NAME OF COUNTRY] 2 Why do we

Community Participation in Research STREAM Clinical Trial [INSERT NAME OF COUNTRY] 2 Why do we need medical research? It is the only way we have of accumulating evidence (proof) to improve treatment of TB To find out if a new

626 views • 18 slides

More recommend