Approximation of empowerment in the continuous domain

Christoph Salge; Cornelius Glackin; D. Polani

doi:10.1142/S0219525912500798

Approximation of empowerment in the continuous domain

Christoph Salge, Cornelius Glackin, D. Polani

Research output: Contribution to journal › Article › peer-review

14 Citations (Scopus)

99 Downloads (Pure)

Abstract

The empowerment formalism offers a goal-independent utility function fully derived from an agent's embodiment. It produces intrinsic motivations which can be used to generate self-organizing behaviours in agents. One obstacle to the application of empowerment in more demanding (esp. continuous) domains is that previous ways of calculating empowerment have been very time consuming and only provided a proof-of-concept. In this paper we present a new approach to efficiently approximate empowerment as a parallel, linear, Gaussian channel capacity problem. We use pendulum balancing to demonstrate this new method, and compare it to earlier approximation methods.

Original language	English
Article number	1250079
Number of pages	29
Journal	Advances in Complex Systems
Volume	16
Issue number	1&2
DOIs	https://doi.org/10.1142/S0219525912500798
Publication status	Published - 2012

Keywords

Goal Independence
Empowerment
Information Theory
Channel Capacity
Perception-Action Loop
Self-Organization
Continuous Domain
Control Theory

Access to Document

10.1142/S0219525912500798

ContiPaper
Electronic version of an article published as Advances in Complex Systems, vol.16, issue 1&2, art.1250079, DOI: 10.1142/S0219525912500798 copyright World Scientific Publishing Company
Accepted author manuscript, 589 KB

Cite this

@article{069f6f5542594bd0b0741011a94d0cd9,

title = "Approximation of empowerment in the continuous domain",

abstract = "The empowerment formalism offers a goal-independent utility function fully derived from an agent's embodiment. It produces intrinsic motivations which can be used to generate self-organizing behaviours in agents. One obstacle to the application of empowerment in more demanding (esp. continuous) domains is that previous ways of calculating empowerment have been very time consuming and only provided a proof-of-concept. In this paper we present a new approach to efficiently approximate empowerment as a parallel, linear, Gaussian channel capacity problem. We use pendulum balancing to demonstrate this new method, and compare it to earlier approximation methods.",

keywords = "Goal Independence, Empowerment, Information Theory, Channel Capacity, Perception-Action Loop, Self-Organization, Continuous Domain, Control Theory",

author = "Christoph Salge and Cornelius Glackin and D. Polani",

year = "2012",

doi = "10.1142/S0219525912500798",

language = "English",

volume = "16",

journal = "Advances in Complex Systems",

issn = "0219-5259",

publisher = "World Scientific Publishing Co. Pte Ltd",

number = "1&2",

}

TY - JOUR

T1 - Approximation of empowerment in the continuous domain

AU - Salge, Christoph

AU - Glackin, Cornelius

AU - Polani, D.

PY - 2012

Y1 - 2012

N2 - The empowerment formalism offers a goal-independent utility function fully derived from an agent's embodiment. It produces intrinsic motivations which can be used to generate self-organizing behaviours in agents. One obstacle to the application of empowerment in more demanding (esp. continuous) domains is that previous ways of calculating empowerment have been very time consuming and only provided a proof-of-concept. In this paper we present a new approach to efficiently approximate empowerment as a parallel, linear, Gaussian channel capacity problem. We use pendulum balancing to demonstrate this new method, and compare it to earlier approximation methods.

AB - The empowerment formalism offers a goal-independent utility function fully derived from an agent's embodiment. It produces intrinsic motivations which can be used to generate self-organizing behaviours in agents. One obstacle to the application of empowerment in more demanding (esp. continuous) domains is that previous ways of calculating empowerment have been very time consuming and only provided a proof-of-concept. In this paper we present a new approach to efficiently approximate empowerment as a parallel, linear, Gaussian channel capacity problem. We use pendulum balancing to demonstrate this new method, and compare it to earlier approximation methods.

KW - Goal Independence

KW - Empowerment

KW - Information Theory

KW - Channel Capacity

KW - Perception-Action Loop

KW - Self-Organization

KW - Continuous Domain

KW - Control Theory

U2 - 10.1142/S0219525912500798

DO - 10.1142/S0219525912500798

M3 - Article

SN - 0219-5259

VL - 16

JO - Advances in Complex Systems

JF - Advances in Complex Systems

IS - 1&2

M1 - 1250079

ER -

Approximation of empowerment in the continuous domain

Abstract

Keywords

Access to Document

Fingerprint

Cite this