a reinforcement learning and synthetic
play

A Reinforcement Learning and Synthetic Data Approach to Mobile - PowerPoint PPT Presentation

Sep 30, 2023 •235 likes •750 views

A Reinforcement Learning and Synthetic Data Approach to Mobile Notification Management Rowan Sutton, Kieran Fraser, Owen Conlan ADAPT Centre, Trinity College Dublin The ADAPT Centre is funded under the SFI Research Centres Programme (Grant

  1. A Reinforcement Learning and Synthetic Data Approach to Mobile Notification Management Rowan Sutton, Kieran Fraser, Owen Conlan ADAPT Centre, Trinity College Dublin The ADAPT Centre is funded under the SFI Research Centres Programme (Grant 13/RC/2106) and is co-funded under the European Regional Development Fund.

  2. Content www.adaptcentre.ie ❖ Motivation ❖ Research Design ❖ Experiment Implementation ❖ Results ❖ Limitations & Future Work ❖ Conclusion

  3. Motivation: Anecdotal www.adaptcentre.ie

  4. Motivation: SOTA www.adaptcentre.ie Growing number of Notification notifications pushed delivery not smart at users (Pielot, M. (Mehrotra, A. et al, et al, 2014). 2016). Unnecessary Large no. of notifications may incoming dramatically notifications = decrease negative user productivity (Iqbal, S. emotions (Sahami T. et al, 2010). Shirazi, A. et al, 2014).

  5. Motivation: Observed Problem www.adaptcentre.ie

  6. Research Design: Gathering Data www.adaptcentre.ie WeAreUs Android App ❖ Experience Sampling Method ❖ Moments of notification interest, moments of phone usage interest ❖ Anonymised & Synthesised

  7. Research Design: Gathering Data www.adaptcentre.ie 31,329 15 participants notifications over 3 months logged 291 4,940 questionnaire smartphone responses usage logs

  8. Research Design: Data Analysis www.adaptcentre.ie

  9. Research Design: Data Analysis www.adaptcentre.ie

  10. Research Design: Data Analysis www.adaptcentre.ie

  11. Research Design: Data Analysis www.adaptcentre.ie

  12. Research Design: Data Analysis www.adaptcentre.ie

  13. Research Design: Synthesising Data www.adaptcentre.ie

  14. Research Design: Synthesising Data www.adaptcentre.ie

  15. Research Design: Synthesising Data www.adaptcentre.ie

  16. Research Design: Synthesising Data www.adaptcentre.ie

  17. Research Design: Synthesising Data www.adaptcentre.ie Train on Real, Test on Synthetic 1 RMSE F1 scores differ in range 0.02 – 0.07 indicating synthetic data imitates real world data. 1. Esteban, C., Hyland, S.L., Ratsch, G.: Real-valued (medical) time series generation

  18. Research Design: Reinforcement Learning www.adaptcentre.ie OpenAI Gym Open source toolkit for “developing and comparing reinforcement learning algorithms” 1 Gym-Push Custom OpenAI Gym environment simulating push-notification overload on mobile device users 1. https://gym.openai.com/

  19. Research Design: Reinforcement Learning www.adaptcentre.ie Gym-Push Custom OpenAI Gym environment simulating push-notification overload on mobile device users State Action Context + Open / Dismiss the Notification Features notification

  20. Experiment Implementation www.adaptcentre.ie Q-learning Agent • Learn a policy to maximise total reward • Create q-table to track quality of state->action pairs • Updates q-values according to Watkins one-step Q-learning algorithm (1) • Can explore or exploit ( ε ) Deep Q-learning Agent • Replaces the q-table with a DNN • Takes the state as input and output is an action • Weights optimised based on the Huber Loss function (2)

  21. Experiment Implementation www.adaptcentre.ie Individual User (Synth & Balanced) • Comprised of ≈ 6000 synthetic notifications • Split into sets of size: 50, 100, 250, 500, 1000, 2500, 5000 • Balanced Individual User (Real & Balanced) • Comprised of ≈ 6000 real notifications • Split into sets of size: 50, 100, 250, 500, 1000, 2500, 5000 • Balanced Individual User (Real & Unbalanced) • Comprised of ≈ 6000 real notifications • Split into sets of size: 50, 100, 250, 500, 1000, 2500, 5000 • Unbalanced Multiple Users (Real & Unbalanced) • Comprised of ≈ 1000 real notifications • Unbalanced

  22. Experiment Implementation www.adaptcentre.ie ❖ Evaluating agents ability to correctly predict user action of open/dismiss notification ❖ Feature set: { app, category, time-of-day, day-of- week } ❖ Evaluate with 10-fold cross validation ❖ Accuracy ❖ Precision – important when cost of false positive is high e.g. agent predicts user wants to see it, delivers -> they end up dismissing it ❖ Recall – important when cost of false negative is high e.g. agent predicts user doesn’t need to see it, caches it -> they miss an important message ❖ F1

  23. Results: Q-learning on synthetic data www.adaptcentre.ie

  24. Results: Q-learning on synthetic data www.adaptcentre.ie

  25. Results: Q-learning on synthetic data www.adaptcentre.ie

  26. Results: Q-learning on real data www.adaptcentre.ie

  27. Results: Q-learning on real data www.adaptcentre.ie

  28. Results: State Space Impact www.adaptcentre.ie Synthetic Data Real Data

  29. Applied Research: Observed User Problem www.adaptcentre.ie

  30. Applied Research: Q-learning Solution www.adaptcentre.ie

  31. Results: DQN on synthetic data www.adaptcentre.ie

  32. Results: DQN on synthetic data www.adaptcentre.ie

  33. Results: DQN on synthetic data www.adaptcentre.ie

  34. Results: DQN on real data www.adaptcentre.ie

  35. Results: DQN on real data www.adaptcentre.ie

  36. Applied Research: Observed User Problem www.adaptcentre.ie

  37. Applied Research: DQN Solution www.adaptcentre.ie

  38. Results: Feature importance www.adaptcentre.ie

  39. Results: Train on Synthetic, Test on Real www.adaptcentre.ie

  40. Results: Train on Synthetic, Test on Real www.adaptcentre.ie

  41. Results: Multiple Users www.adaptcentre.ie

  42. Results: Multiple Users www.adaptcentre.ie

  43. Limitations & Future Work www.adaptcentre.ie Limitations • Small set of users • Restricted set of features e.g. ticker text not used • Fixed hyper-parameters Future Work • Generative modeling applied to text • Exploring other RL algorithms e.g. HER, IMPALA • Larger user study

  44. Future Work – Conditional Ticker Text Generation www.adaptcentre.ie

  45. Future Work – Autonomous Personalised Notifications www.adaptcentre.ie

  46. Future Work – Autonomous Personalised Notifications www.adaptcentre.ie

  47. Future Work – Autonomous Personalised Notifications www.adaptcentre.ie

  48. Conclusion www.adaptcentre.ie OpenAI Gym Shareable notification environment for training on data set notifications Two methods of RL Evaluations illustrate applied to agents achieve notification comparable management performance to SOTA

  49. EvalUMAP www.adaptcentre.ie http://evalumap.adaptcentre.ie/

  50. www.adaptcentre.ie Thank you. Questions? Demo: https://review2019.github.io Email: kieran.fraser@adaptcentre.ie

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
Presentation contents 1. Role of geosynthetic veneer reinforcement in barrier systems 2.

Presentation contents 1. Role of geosynthetic veneer reinforcement in barrier systems 2.

Landfill 2015 Geogrid reinforcement in harsh environments Peter er Assinde der HUESKER Synthetic GmbH, Manager AFRICA Laura a Ca Carbone ne HUESKER Synthetic GmbH, Engineering department Morne ne Breyten ytenbach bach HUESKER

357 views • 24 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
Jugger - 5pm Thursdays on Memorial Glade cs160. cs160. valkyriesavage.com valkyriesavage.com

Jugger - 5pm Thursdays on Memorial Glade cs160. cs160. valkyriesavage.com valkyriesavage.com

Jugger - 5pm Thursdays on Memorial Glade cs160. cs160. valkyriesavage.com valkyriesavage.com task analysis & contextual inquiry July 01, 2015 Valkyrie Savage Y ou can observe a lot just by watching. Yogi Berra main points of

1.03k views • 81 slides
Creating a recruitment plan FOR your participants, not just for your research Ashley Smith,

Creating a recruitment plan FOR your participants, not just for your research Ashley Smith,

RPN Workshop: November 21, 2019 Putting Participants First: Creating a recruitment plan FOR your participants, not just for your research Ashley Smith, Boston Medical Center | Deaven Hough, University of Florida Heres what we got going on

562 views • 45 slides
Dimension-Wise Importance Sampling Weight Clipping for Sample-Efficient Reinforcement Learning

Dimension-Wise Importance Sampling Weight Clipping for Sample-Efficient Reinforcement Learning

Han and Sung, ICML 2019 c 1 Dimension-Wise Importance Sampling Weight Clipping for Sample-Efficient Reinforcement Learning Seungyul Han and Youngchul Sung Dept. of Electrical Engineering KAIST ICML 2019, Long Beach, CA, USA Jun. 12,

275 views • 10 slides
Improving Accuracy in End-to-end Packet Loss Measurement * Joel Sommers, Paul Barford, Nick

Improving Accuracy in End-to-end Packet Loss Measurement * Joel Sommers, Paul Barford, Nick

Improving Accuracy in End-to-end Packet Loss Measurement * Joel Sommers, Paul Barford, Nick Duffield, Amos Ron University of Wisconsin-Madison * AT&T Labs-Research Background Understanding its basic characteristics is important

344 views • 22 slides
Notes on Quantitative UX Research at Google Chris Chapman Quantitative UX Researcher Overview

Notes on Quantitative UX Research at Google Chris Chapman Quantitative UX Researcher Overview

Notes on Quantitative UX Research at Google Chris Chapman Quantitative UX Researcher Overview This is a personal, unofficial view on Quantitative UX research at Google. I share opinions about the job, who may enjoy it, reflections on hiring,

339 views • 5 slides
Unified View width of backup Dynamic Temporal- programming difference learning height

Unified View width of backup Dynamic Temporal- programming difference learning height

Unified View width of backup Dynamic Temporal- programming difference learning height (depth) of backup Exhaustive Monte search Carlo ... R. S. Sutton and A. G. Barto: Reinforcement Learning: An Introduction 1 Chapter 8: Planning

538 views • 24 slides
Chapter 5: Monte Carlo Methods Monte Carlo methods are learning methods Experience

Chapter 5: Monte Carlo Methods Monte Carlo methods are learning methods Experience

Chapter 5: Monte Carlo Methods Monte Carlo methods are learning methods Experience values, policy Monte Carlo methods can be used in two ways: ! model-free: No model necessary and still attains optimality ! Simulated: Needs only a

2.16k views • 32 slides
Monte Carlo Methods CS60077: Reinforcement Learning Abir Das IIT Kharagpur Sep 06 and 12, 2019

Monte Carlo Methods CS60077: Reinforcement Learning Abir Das IIT Kharagpur Sep 06 and 12, 2019

Monte Carlo Methods CS60077: Reinforcement Learning Abir Das IIT Kharagpur Sep 06 and 12, 2019 Agenda Introduction MC Evaluation MC Control Agenda Understand how to evaluate policies in model-free setting using Monte Carlo methods

751 views • 53 slides

More recommend