Dynamics of Learning & Distributed Adaptation

1
Dynamics of Learning & Distributed Adaptation Santa Fe Institute: James P. Crutchfield, P.I. Future Plans (6 months out) New problems: Continuous-state and continuous-time agents Adaptation to active, pattern-forming environments Dynamical theory of how learning and adaptation occur Anticipated results: Monitor emergence of cooperation in agent collectives Measure mutuality in interacting reinforcement learne Test on in-house autonomous robotic vehicle collectiv Analytical tools: Predict whether or not group cooperation can occur Agent intelligence versus group size Prediction of the rate of adaptation during collectiv Prototype models: Solvable MAS systems Software tools: Ab Initio Learning Algorithms Library for Estimating MASS Metrics Enterprise Java Platform for Robot Collectives Multi-Agent System Science (MASS) Dimension Agents learn complex environment ab initio Synchronization of agent to environment Agents adapt to nonstationary environment Strategies for agent-agent coordination Metrics for large-scale MASs Statistical Complexity: Amount of structure & organization in environ’t Individual agent knowledge v. group knowledge Mutuality: Architecture of information flow Lyapunov Spectra: Degrees of stability and instability Causal Synchrony: Detect coherent subgroup behavior CAHDE REF ACFC: Adapting to instabilities in air flow control AirOps: Emergence of spontaneous leadership Solution: Interacting reinforcement and -machine learning agents solve a group task Approach: Pattern Discovery: Beyond pattern recognition Design & analysis based on sound principles of learning Metrics for cooperation in large-scale systems Results To Date Predictive theory of agent learning: Quantify agent modeling capacity Data Set Size v. Prediction Error v. Model Complexity Pattern Discovery: The “Aha” Effect Incremental learning algorithm Quantify structure in environment: How structure leads to unpredictability for agent Define synchronization for chaotic environments: Predict required data and time to synchronize Periodic case solved in closed form Transient information: New metric of synchronization

description

Dynamics of Learning & Distributed Adaptation. Santa Fe Institute: James P. Crutchfield, P.I. Multi-Agent System Science (MASS) Dimension Agents learn complex environment ab initio Synchronization of agent to environment Agents adapt to nonstationary environment - PowerPoint PPT Presentation

Transcript of Dynamics of Learning & Distributed Adaptation

Page 1: Dynamics of Learning & Distributed Adaptation

Dynamics of Learning & Distributed Adaptation Santa Fe Institute: James P. Crutchfield, P.I.

Future Plans (6 months out)New problems:

Continuous-state and continuous-time agentsAdaptation to active, pattern-forming environmentsDynamical theory of how learning and adaptation occur

Anticipated results:Monitor emergence of cooperation in agent collectivesMeasure mutuality in interacting reinforcement learnersTest on in-house autonomous robotic vehicle collectives

Analytical tools:Predict whether or not group cooperation can occurAgent intelligence versus group sizePrediction of the rate of adaptation during collective taskPrototype models: Solvable MAS systems

Software tools:Ab Initio Learning AlgorithmsLibrary for Estimating MASS MetricsEnterprise Java Platform for Robot Collectives

Multi-Agent System Science (MASS)Dimension

Agents learn complex environment ab initioSynchronization of agent to environmentAgents adapt to nonstationary environmentStrategies for agent-agent coordination

Metrics for large-scale MASsStatistical Complexity:

Amount of structure & organization in environ’tIndividual agent knowledge v. group knowledge

Mutuality: Architecture of information flowLyapunov Spectra: Degrees of stability and instabilityCausal Synchrony: Detect coherent subgroup behavior

CAHDE REF

ACFC: Adapting to instabilities in air flow controlAirOps: Emergence of spontaneous leadership

Solution:Interacting reinforcement and -machine learning agents solve a group task

Approach:

Pattern Discovery: Beyond pattern recognition

Design & analysis based on sound principles of learning

Metrics for cooperation in large-scale systemsResults To DatePredictive theory of agent learning:

Quantify agent modeling capacity

Data Set Size v. Prediction Error v. Model Complexity

Pattern Discovery: The “Aha” Effect

Incremental learning algorithm

Quantify structure in environment:

How structure leads to unpredictability for agent

Define synchronization for chaotic environments:

Predict required data and time to synchronize

Periodic case solved in closed form

Transient information: New metric of synchronization

Dynamics of reinforcement-learning agents:

Nash equilibria v. oscillation v. chaos

Dependence on system architecture and initial state