Behavior in its general form can be defined as a mapping between sensory inputs and a pattern of motor actions that are used to achieve a goal. Reinforcement learning in the last years emerged as a general framework to analyze behavior in its general definition. In this thesis exploiting the techniques of reinforcement learning we study several phenomena that can be classified as search, navigation and foraging behaviors. Regarding the search aspect we analyze random walks forced to reach a target in a confined region of the space. In this case we can solve analytically the problem that allows to find a very efficient way to generate such walks. The navigation problem is inspired by olfactory navigation in homing pigeons. In this case we propose an algorithm to navigate a noisy environment relying only on local signals. The foraging instead is analyzed starting from the observation that fossil traces show the evolution of foraging strategies towards highly compact and self-avoiding trajectories. We show how this optimal behavior can emerge in the reinforcement learning framework.

Search, navigation and foraging: an optimal decision-making perspective / Adorisio, Matteo. - (2018 Oct 29).

Search, navigation and foraging: an optimal decision-making perspective

Adorisio, Matteo
2018-10-29

Abstract

Behavior in its general form can be defined as a mapping between sensory inputs and a pattern of motor actions that are used to achieve a goal. Reinforcement learning in the last years emerged as a general framework to analyze behavior in its general definition. In this thesis exploiting the techniques of reinforcement learning we study several phenomena that can be classified as search, navigation and foraging behaviors. Regarding the search aspect we analyze random walks forced to reach a target in a confined region of the space. In this case we can solve analytically the problem that allows to find a very efficient way to generate such walks. The navigation problem is inspired by olfactory navigation in homing pigeons. In this case we propose an algorithm to navigate a noisy environment relying only on local signals. The foraging instead is analyzed starting from the observation that fossil traces show the evolution of foraging strategies towards highly compact and self-avoiding trajectories. We show how this optimal behavior can emerge in the reinforcement learning framework.
Micheletti, Cristian
Celani, Antonio
Adorisio, Matteo
File in questo prodotto:
File Dimensione Formato  
main.pdf

accesso aperto

Descrizione: Tesi di PhD
Tipologia: Tesi
Licenza: Non specificato
Dimensione 11.24 MB
Formato Adobe PDF
11.24 MB Adobe PDF Visualizza/Apri

I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.

Utilizza questo identificativo per citare o creare un link a questo documento: http://hdl.handle.net/20.500.11767/84102
Citazioni
  • ???jsp.display-item.citation.pmc??? ND
  • Scopus ND
  • ???jsp.display-item.citation.isi??? ND
social impact