### Abstract

One of the fundamental problems of an interconnected interactive system is the huge amounts of data that are being generated by every entity. Unfortunately, what we seek is not data but information, and therefore, a growing bottleneck is exactly how to extract and learn useful information from data. In this paper, the information-Theoretic learning in data-driven games is studied. This learning shows that the imitative Boltzmann-Gibbs strategy is the maximizer of the perturbed payoff where the perturbation function is the relative entropy from the previous strategy to the current one. In particular, the imitative strategy is the best learning scheme with the respect to data-driven games with cost of moves. Based on it, the classical imitative Boltzmann-Gibbs learning in data-driven games is revisited. Due to communication complexity and noisy data measurements, the classical imitative Boltzmann-Gibbs cannot be applied directly in situations were only numerical values of player's own payoff is measured. A combined fully distributed payoff and strategy imitative learning (CODIPAS) is proposed. Connections between the rest points of the resulting game dynamics, equilibria are established.

Original language | English (US) |
---|---|

Title of host publication | Proceedings of 2017 IEEE 6th Data Driven Control and Learning Systems Conference, DDCLS 2017 |

Publisher | Institute of Electrical and Electronics Engineers Inc. |

Pages | 22-29 |

Number of pages | 8 |

ISBN (Electronic) | 9781509054619 |

DOIs | |

State | Published - Oct 13 2017 |

Event | 6th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2017 - Chongqing, China Duration: May 26 2017 → May 27 2017 |

### Other

Other | 6th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2017 |
---|---|

Country | China |

City | Chongqing |

Period | 5/26/17 → 5/27/17 |

### Fingerprint

### Keywords

- Data-driven Learning
- Distributed Systems
- Game Dynamics
- Imitation
- Noisy Games

### ASJC Scopus subject areas

- Computational Mechanics
- Artificial Intelligence
- Computer Networks and Communications

### Cite this

*Proceedings of 2017 IEEE 6th Data Driven Control and Learning Systems Conference, DDCLS 2017*(pp. 22-29). [8067719] Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/DDCLS.2017.8067719

**Data-driven vs model-driven imitative learning.** / Hamidou, Tembine.

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution

*Proceedings of 2017 IEEE 6th Data Driven Control and Learning Systems Conference, DDCLS 2017.*, 8067719, Institute of Electrical and Electronics Engineers Inc., pp. 22-29, 6th IEEE Data Driven Control and Learning Systems Conference, DDCLS 2017, Chongqing, China, 5/26/17. https://doi.org/10.1109/DDCLS.2017.8067719

}

TY - GEN

T1 - Data-driven vs model-driven imitative learning

AU - Hamidou, Tembine

PY - 2017/10/13

Y1 - 2017/10/13

N2 - One of the fundamental problems of an interconnected interactive system is the huge amounts of data that are being generated by every entity. Unfortunately, what we seek is not data but information, and therefore, a growing bottleneck is exactly how to extract and learn useful information from data. In this paper, the information-Theoretic learning in data-driven games is studied. This learning shows that the imitative Boltzmann-Gibbs strategy is the maximizer of the perturbed payoff where the perturbation function is the relative entropy from the previous strategy to the current one. In particular, the imitative strategy is the best learning scheme with the respect to data-driven games with cost of moves. Based on it, the classical imitative Boltzmann-Gibbs learning in data-driven games is revisited. Due to communication complexity and noisy data measurements, the classical imitative Boltzmann-Gibbs cannot be applied directly in situations were only numerical values of player's own payoff is measured. A combined fully distributed payoff and strategy imitative learning (CODIPAS) is proposed. Connections between the rest points of the resulting game dynamics, equilibria are established.

AB - One of the fundamental problems of an interconnected interactive system is the huge amounts of data that are being generated by every entity. Unfortunately, what we seek is not data but information, and therefore, a growing bottleneck is exactly how to extract and learn useful information from data. In this paper, the information-Theoretic learning in data-driven games is studied. This learning shows that the imitative Boltzmann-Gibbs strategy is the maximizer of the perturbed payoff where the perturbation function is the relative entropy from the previous strategy to the current one. In particular, the imitative strategy is the best learning scheme with the respect to data-driven games with cost of moves. Based on it, the classical imitative Boltzmann-Gibbs learning in data-driven games is revisited. Due to communication complexity and noisy data measurements, the classical imitative Boltzmann-Gibbs cannot be applied directly in situations were only numerical values of player's own payoff is measured. A combined fully distributed payoff and strategy imitative learning (CODIPAS) is proposed. Connections between the rest points of the resulting game dynamics, equilibria are established.

KW - Data-driven Learning

KW - Distributed Systems

KW - Game Dynamics

KW - Imitation

KW - Noisy Games

UR - http://www.scopus.com/inward/record.url?scp=85034033934&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85034033934&partnerID=8YFLogxK

U2 - 10.1109/DDCLS.2017.8067719

DO - 10.1109/DDCLS.2017.8067719

M3 - Conference contribution

SP - 22

EP - 29

BT - Proceedings of 2017 IEEE 6th Data Driven Control and Learning Systems Conference, DDCLS 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -