on learning to think algorithmic information theory for
play

On Learning to Think: Algorithmic Information Theory for Novel - PowerPoint PPT Presentation

Oct 13, 2023 •361 likes •537 views

On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models Jrgen Schmidhuber The Swiss AI Lab IDSIA Univ. Lugano & SUPSI

  1. On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models Jürgen Schmidhuber The Swiss AI Lab IDSIA Univ. Lugano & SUPSI http://www.idsia.ch/~juergen NNAISENSE

  2. Jürgen Schmidhuber You_again Shmidhoobuh

  3. http://www.idsia.ch/~juergen/rnn.html 1997-2009. Since 2015 on your phone! Google, Microsoft, IBM, Apple, all use LSTM now With Hochreiter (1997), Gers (2000), Graves, Fernandez, Gomez, Bayer…

  4. LSTM learns knot-tying tasklets: Mayr Gomez Wierstra Nagy Knoll Schmidhuber, IROS’06

  5. 2005: Reinforcement- Learning or Evolving RNNs with Fast Weights Robot learns to balance 1 or 2 poles through 3D joint Gomez & Schmidhuber: Co-evolving recurrent neurons learn deep memory POMDPs. GECCO 2005 http://www.idsia.ch/~juergen/evolution.html

  6. Reinforcement Learning in Partially Observable Worlds Finds Complex Neural Controllers with a Million Weights – RAW VIDEO INPUT! Faustino Gomez, Jan Koutnik, Giuseppe Cuccu, J. Schmidhuber, GECCO, July 2013

  7. J.S.: IJCNN 1990, NIPS 1991: Reinforcement Learning with Recurrent Controller & Recurrent World Model Learning and planning with recurrent networks

  8. IJNS 1991: R-Learning of Visual Attention on 100,000 times slower computers http://people.idsia.ch/~juergen/attentive.html

  9. 1991: current goal=extra fixed input 2015: all of this is coming back!

  10. RoboCup World Champion 2004, Fastest League, 5m/s Lookahead expectation & planning with neural networks (Schmidhuber, IEEE INNS 1990): successfully used for RoboCup by Alexander Gloye-Förster (went to IDSIA) http://www.idsia.ch/~juergen/learningrobots.html Alex @ IDSIA, led FU Berlin’s RoboCup World Champion Team 2004

  11. RNNAIssance 2014-2015 On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning RNN- based Controllers (RNNAIs) and Recurrent Neural World Models http://arxiv.org/abs/1511.09249

  12. Maximize Future Fun(Data X,O(t))~ ∂ CompResources(X,O(t))/ ∂ t Formal theory of fun & novelty & surprise & attention & creativity & curiosity & art & science & humor E.g., Connection Science 18(2):173-187, 2006 IEEE Transactions AMD 2(3):230-247, 2010 http://www.idsia.ch/~juergen/creativity.html

  13. https://www.youtube.com/watch?v=OTqdXbTEZpE Continual curiosity-driven skill acquisition from high-dimensional video inputs for humanoid robots. Kompella, Stollenga, Luciw, Schmidhuber. Artificial Intelligence, 2015

  14. neural networks-based artificial intelligence now talking to investors

  15. NIPS 2016 demo: Reinforcement learning to park Cooperation NNAISENSE - AUDI

Recommend

Dualities and Dichotomies in Algorithmic Information Theory Jan Reimann Pennsylvania State

Dualities and Dichotomies in Algorithmic Information Theory Jan Reimann Pennsylvania State

Dualities and Dichotomies in Algorithmic Information Theory Jan Reimann Pennsylvania State University July 15, 2011 Algorithmic Information Theory brings together information theory and computability theory; develops a framework

735 views • 36 slides
SALZA: Algorithmic Information Theory and Universal Classification for Sequences SeqBio 2018,

SALZA: Algorithmic Information Theory and Universal Classification for Sequences SeqBio 2018,

General information measures SALZA similarity as information Applications of SALZA Parallel implementation SALZA: Algorithmic Information Theory and Universal Classification for Sequences SeqBio 2018, Rouen, France Franois Cayre, Nicolas Le

719 views • 43 slides
Bo Waggoner ? Harvard Advisor: Yiling Chen Main area: algorithmic game theory Specialty:

Bo Waggoner ? Harvard Advisor: Yiling Chen Main area: algorithmic game theory Specialty:

Bo Waggoner ? Harvard Advisor: Yiling Chen Main area: algorithmic game theory Specialty: foundations of information and incentives ? Other interests: online and randomized algorithms, distribution learning/testing, . (always looking to

55 views • 3 slides
An application of algorithmic information theory Jennifer Chubb George Washington University

An application of algorithmic information theory Jennifer Chubb George Washington University

An application of algorithmic information theory Jennifer Chubb George Washington University Washington, DC Graduate Student Seminar March 9, 2007 Slides available at home.gwu.edu/ jchubb Introduction Initial segments of scattered

310 views • 13 slides
Algorithmic Decision Theory and Smart Cities Fred Roberts Rutgers University 1 Algorithmic

Algorithmic Decision Theory and Smart Cities Fred Roberts Rutgers University 1 Algorithmic

Algorithmic Decision Theory and Smart Cities Fred Roberts Rutgers University 1 Algorithmic Decision Theory Todays decision makers in fields ranging from engineering to medicine to homeland security have available to them:

508 views • 34 slides
Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part II: Algorithmic Applications of WQOs

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part II: Algorithmic Applications of WQOs

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part II: Algorithmic Applications of WQOs Sylvain Schmitz & Philippe Schnoebelen LSV, CNRS & ENS Cachan ESSLLI 2016, Bozen/Bolzano, Aug 22-26, 2016 Lecture notes & exercises

312 views • 30 slides
Algorithmic Learning Theory Theoretical Computer Science Peter Rossmanith Felix Reidl, Fernando

Algorithmic Learning Theory Theoretical Computer Science Peter Rossmanith Felix Reidl, Fernando

Algorithmic Learning Theory Theoretical Computer Science Peter Rossmanith Felix Reidl, Fernando Sanchez, Somnath Sikdar Ground Rules 1. A talk each week (venue to be decided). 45 mins duration + 45 mins for discussion in German or

388 views • 10 slides
PAC Learning prof. dr Arno Siebes Algorithmic Data Analysis Group Department of Information and

PAC Learning prof. dr Arno Siebes Algorithmic Data Analysis Group Department of Information and

PAC Learning prof. dr Arno Siebes Algorithmic Data Analysis Group Department of Information and Computing Sciences Universiteit Utrecht Recall: PAC Learning (Version 1) A hypothesis class H is PAC learnable if there exists a function m H : (0 ,

705 views • 29 slides
Information, Learning and Falsification David Balduzzi December 17, 2011 Max Planck Institute

Information, Learning and Falsification David Balduzzi December 17, 2011 Max Planck Institute

Effective information Algorithmic information Learning theory Falsification Conclusion Information, Learning and Falsification David Balduzzi December 17, 2011 Max Planck Institute for Intelligent Systems T ubingen, Germany Effective

695 views • 43 slides
Computational Learning Theory 1 / 22 Decidability Computation Decidability which

Computational Learning Theory 1 / 22 Decidability Computation Decidability which

Computational Learning Theory 1 / 22 Decidability Computation Decidability which problems have algorithmic solutions Machine Learning Feasibility what assumptions must we make to trust that we can learn an unknown target

246 views • 22 slides
T-61.182 Information Theory and Machine Learning 38. Introduction to Neural Networks 40.

T-61.182 Information Theory and Machine Learning 38. Introduction to Neural Networks 40.

T-61.182 Information Theory and Machine Learning 38. Introduction to Neural Networks 40. Capacity of a Single Neuron 41. Learning as Inference Presented by Yang, Zhi-rong on 22, April 2004 T-61.182 Information Theory and Machine Learning

423 views • 20 slides
Extensive Form Games Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC Perfect

Extensive Form Games Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC Perfect

Extensive Form Games Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC Perfect Information Extensive Form Games Backward Induction and MinMax Algorithms Imperfect Information Extensive Form Games The Sequence Form

521 views • 32 slides
Point-free Descriptive Set Theory and Algorithmic Randomness Alex Simpson University of

Point-free Descriptive Set Theory and Algorithmic Randomness Alex Simpson University of

1/24 Point-free Descriptive Set Theory and Algorithmic Randomness Alex Simpson University of Ljubljana, Slovenia incorporating j.w.w. Antonin Delpeuch (Univ. Oxford) CCC, Nancy, June 2017 Point-free Descriptive Set Theory and Algorithmic

538 views • 24 slides
Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain Schmitz & Philippe Schnoebelen LSV, CNRS & ENS Cachan ESSLLI 2016, Bozen/Bolzano, Aug 22-26, 2016 Lecture notes & exercises available at

592 views • 58 slides
Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain

Algorithmic Aspects of WQO (Well-Quasi-Ordering) Theory Part I: Basics of WQO Theory Sylvain Schmitz & Philippe Schnoebelen LSV, CNRS & ENS Cachan ESSLLI 2012, Opole, Aug 6-15, 2012 Lecture notes & exercices available at

536 views • 50 slides
max ( | ) ( ) P s a U s preferences, must exist consistent utility function a s

max ( | ) ( ) P s a U s preferences, must exist consistent utility function a s

The Winding Path to RL Markov Decision Processes & Decision Theory Descriptive theory of optimal Reinforcement Learning behavior Markov Decision Processes Mathematical/Algorithmic realization of Decision Theory

86 views • 8 slides
Fractal Intersections and Products via Algorithmic Dimension Neil Lutz Rutgers University June

Fractal Intersections and Products via Algorithmic Dimension Neil Lutz Rutgers University June

Fractal Intersections and Products via Algorithmic Dimension Neil Lutz Rutgers University June 26, 2017 Goal: Use algorithmic information theory to answer fundamental questions in fractal geometry. Agenda: Discuss classical and

1.27k views • 80 slides
CS480/680 Machine Learning Lecture 5: January 21 st , 2020 Information Theory Zahra Sheikhbahaee

CS480/680 Machine Learning Lecture 5: January 21 st , 2020 Information Theory Zahra Sheikhbahaee

CS480/680 Machine Learning Lecture 5: January 21 st , 2020 Information Theory Zahra Sheikhbahaee Sources: Elements Of Information Theory Information Theory, Inference, and Learning Algorithms University of Waterloo CS480/680 Winter 2020 Zahra

339 views • 16 slides
Basic Algorithms in Number Theory Francesco Pappalardi #1 - Algorithmic Complexity & more.

Basic Algorithms in Number Theory Francesco Pappalardi #1 - Algorithmic Complexity & more.

Algorithmic Complexity ... 1 Basic Algorithms in Number Theory Basic Algorithms in Number Theory Francesco Pappalardi #1 - Algorithmic Complexity & more. August 31 st 2015 SEAMS School 2015 Number Theory and Applications in Cryptography

723 views • 27 slides
Coalitional Game Theory Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC

Coalitional Game Theory Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC

Coalitional Game Theory Game Theory MohammadAmin Fazli Algorithmic Game Theory 1 TOC Coalitional Games Fair Division and Shapley Value Stable Division and the Core Concept -Core, Least core & Nucleolus Reading:

422 views • 24 slides
Chapter 1: Introduction to Information Theory Book: Information Theory, Inference, and

Chapter 1: Introduction to Information Theory Book: Information Theory, Inference, and

Chapter 1: Introduction to Information Theory Book: Information Theory, Inference, and Learning Algorithms from David MacKay Noisy Channels - Error correcting Codes Examples - System solution - Channel models - Binary symmetric Channel

571 views • 22 slides
Pattern Languages Seminar Algorithmic Learning Theory, SS 2015 Michael Krause RWTH Aachen

Pattern Languages Seminar Algorithmic Learning Theory, SS 2015 Michael Krause RWTH Aachen

Pattern Languages Seminar Algorithmic Learning Theory, SS 2015 Michael Krause RWTH Aachen 07.05.2015 Michael Krause (RWTH Aachen) Pattern Languages 07.05.2015 1 / 44 Basic ideas 1 Finding Patterns Common to a Set of Strings 2 Other

1.65k views • 148 slides
Algorithmic Game Theory: Introduction Georgios Amanatidis amanatidis@diag.uniroma1.it Based on

Algorithmic Game Theory: Introduction Georgios Amanatidis amanatidis@diag.uniroma1.it Based on

Algorithmic Game Theory: Introduction Georgios Amanatidis amanatidis@diag.uniroma1.it Based on slides by Vangelis Markakis What is Game Theory? Quote from M. Osborne: Game Theory aims to help us understand situations in which decision makers

834 views • 36 slides
Algorithmic number theory and the allied theory of theta functions Christophe Ritzenthaler

Algorithmic number theory and the allied theory of theta functions Christophe Ritzenthaler

Algorithmic number theory and the allied theory of theta functions Christophe Ritzenthaler Institut de Mathmatiques de Luminy, CNRS Edinburgh 10-10 e-mail: ritzenth@iml.univ-mrs.fr web: http://iml.univ-mrs.fr/ ritzenth/ Christophe

569 views • 37 slides

More recommend