東華大學圖書館 |

Language: English

Help

回圖書館首頁

手機版館藏查詢

Back

Switch To: Labeled | MARC Mode | ISBD

Explaining Collective Behavior with ...

Shams, Daniel .

Linked to FindBook

Google Book

Amazon

博客來

Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning.

Record Type:	Electronic resources : Monograph/item
Title/Author:	Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning./
Author:	Shams, Daniel .
Published:	Ann Arbor : ProQuest Dissertations & Theses, : 2019,
Description:	130 p.
Notes:	Source: Dissertations Abstracts International, Volume: 81-08, Section: B.
Contained By:	Dissertations Abstracts International81-08B.
Subject:	Biology. -
Online resource:	http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=27541731
ISBN:	9781392739846

Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning.
Shams, Daniel .

Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning. - Ann Arbor : ProQuest Dissertations & Theses, 2019 - 130 p.

Source: Dissertations Abstracts International, Volume: 81-08, Section: B.

Thesis (Ph.D.)--Northwestern University, 2019.

This item must not be sold to any third party vendors.

Collective behavior exists in countless forms. Two forms of collective behavior from wildly different fields of study will be analyzed in this work. First, the social amoeba Dictyostelium discoideum is discussed. D.discoideum performs collective chemotaxis under starvation conditions where single cells aggregate towards clusters of cells following waves of the signaling molecule cAMP to create a multicellular fruiting body. Taking inspiration from D. discoideum a model is proposed that directly links the relaying of a chemical message to the directional sensing of that signal. Utilizing an excitable dynamical systems model that has been previously validated experimentally, it's possible to have both signal amplification and perfect adaptation in a single module. Through this model, it can be shown that noise plays a vital role in chemotaxis and that the internal timescale of adaptation of the model automatically matches the periodicity of the traveling chemical waves generated in the population. Finally, the interplay of noise and diffusion in creating dynamical instabilities that impede chemotactic ability in a continuum version of the model is discussed.The second form of collective behavior presented is departure from biology and a migration to computer learning. Despite notable progress in recent years, multiagent reinforcement learning is still in its infancy. Many of the theoretical guarantees that apply in a single-agent setting no longer applies when there are more than one learning agent. By adapting methods from dynamical systems theory and applying them to simple game theory settings, a framework is provided for better understanding the agents' behavior. Through this framework, common reinforcement learning algorithms such as Q-learning exhibit behavioral characteristics that can be analytically mapped to concepts in game theory such as the mixed strategy Nash equilibrium. This consideration can lead to a more comprehensive understanding of agent behaviors in terms of fixed points and basins of attraction.

ISBN: 9781392739846Subjects--Topical Terms:

522710
Biology.
Subjects--Index Terms:

Collective behavior

Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning.
LDR:03245nmm a2200349 4500 001 2272871
005 20201105110246.5
008 220629s2019 ||||||||||||||||| ||eng d
020 $a 9781392739846
035 $a (MiAaPQ)AAI27541731
035 $a AAI27541731
040 $a MiAaPQ $c MiAaPQ
100 1 $a Shams, Daniel . $3 3550295
245 1 0 $a Explaining Collective Behavior with Dynamical Systems: Spatial Gradient Sensing in Eukaryotic Chemotaxis and Learning Dynamics in Multiagent Reinforcement Learning.
260 1 $a Ann Arbor : $b ProQuest Dissertations & Theses, $c 2019
300 $a 130 p.
500 $a Source: Dissertations Abstracts International, Volume: 81-08, Section: B.
500 $a Advisor: Schwab, David J.
502 $a Thesis (Ph.D.)--Northwestern University, 2019.
506 $a This item must not be sold to any third party vendors.
520 $a Collective behavior exists in countless forms. Two forms of collective behavior from wildly different fields of study will be analyzed in this work. First, the social amoeba Dictyostelium discoideum is discussed. D.discoideum performs collective chemotaxis under starvation conditions where single cells aggregate towards clusters of cells following waves of the signaling molecule cAMP to create a multicellular fruiting body. Taking inspiration from D. discoideum a model is proposed that directly links the relaying of a chemical message to the directional sensing of that signal. Utilizing an excitable dynamical systems model that has been previously validated experimentally, it's possible to have both signal amplification and perfect adaptation in a single module. Through this model, it can be shown that noise plays a vital role in chemotaxis and that the internal timescale of adaptation of the model automatically matches the periodicity of the traveling chemical waves generated in the population. Finally, the interplay of noise and diffusion in creating dynamical instabilities that impede chemotactic ability in a continuum version of the model is discussed.The second form of collective behavior presented is departure from biology and a migration to computer learning. Despite notable progress in recent years, multiagent reinforcement learning is still in its infancy. Many of the theoretical guarantees that apply in a single-agent setting no longer applies when there are more than one learning agent. By adapting methods from dynamical systems theory and applying them to simple game theory settings, a framework is provided for better understanding the agents' behavior. Through this framework, common reinforcement learning algorithms such as Q-learning exhibit behavioral characteristics that can be analytically mapped to concepts in game theory such as the mixed strategy Nash equilibrium. This consideration can lead to a more comprehensive understanding of agent behaviors in terms of fixed points and basins of attraction.
590 $a School code: 0163.
650 4 $a Biology. $3 522710
650 4 $a Biophysics. $3 518360
650 4 $a Artificial intelligence. $3 516317
653 $a Collective behavior
653 $a Social amoebae
653 $a Reinforcement learning
690 $a 0306
690 $a 0786
690 $a 0800
710 2 $a Northwestern University. $b Interdepartmental Biological Sciences (IBiS) Graduate Program. $3 3288393
773 0 $t Dissertations Abstracts International $g 81-08B.
790 $a 0163
791 $a Ph.D.
792 $a 2019
793 $a English
856 4 0 $u http://pqdd.sinica.edu.tw/twdaoapp/servlet/advanced?query=27541731