This research was phase one of an experiment testing three to five day old chicks. The agents’ natural behavior was observed in the order and speed in which they ate an array of food scattered around a maze-like environment. The goal was to discover what strategy the chicks were using to decide the order in which they ate the pieces of food, and how often their performance lined up with different foraging strategies. A total of 35 chicks between the ages of three to five days were tested, both male and female. Each chick was tested through three different versions of a maze made of 7 X 7 boxes totaling an area of 71.4cm X 71.4cm. These experiments were run over the span of three months, with three to five chicks being tested each week for three different mazes. The findings were then compared to the mathematical calculations of a Machine Learning Model in relation to the Nearest Neighbor Principle and the Optimal Strategy. The results found that the chicks followed the Nearest Neighbor principle in their decisions 68% of the time and the Optimal strategy 62% of the time. The overall goal of this full experiment is to use this biological data to create an accurate AI version of a chick for the Ecotwin virtual ecosystem. This accurate ecosystem model will allow the consequences of human intervention to be analyzed in a safe space first before affecting a real ecosystem to promote ecological conservation.
This research was phase one of an experiment testing three to five day old chicks. The agents’ natural behavior was observed in the order and speed in which they ate an array of food scattered around a maze-like environment. The goal was to discover what strategy the chicks were using to decide the order in which they ate the pieces of food, and how often their performance lined up with different foraging strategies. A total of 35 chicks between the ages of three to five days were tested, both male and female. Each chick was tested through three different versions of a maze made of 7 X 7 boxes totaling an area of 71.4cm X 71.4cm. These experiments were run over the span of three months, with three to five chicks being tested each week for three different mazes. The findings were then compared to the mathematical calculations of a Machine Learning Model in relation to the Nearest Neighbor Principle and the Optimal Strategy. The results found that the chicks followed the Nearest Neighbor principle in their decisions 68% of the time and the Optimal strategy 62% of the time. The overall goal of this full experiment is to use this biological data to create an accurate AI version of a chick for the Ecotwin virtual ecosystem. This accurate ecosystem model will allow the consequences of human intervention to be analyzed in a safe space first before affecting a real ecosystem to promote ecological conservation.
A Pilot Experiment to Observe Domestic Chicks’ Foraging Strategies in Comparison with a Machine Learning Model
DAUGHETY, OLIVIA GRACE
2023/2024
Abstract
This research was phase one of an experiment testing three to five day old chicks. The agents’ natural behavior was observed in the order and speed in which they ate an array of food scattered around a maze-like environment. The goal was to discover what strategy the chicks were using to decide the order in which they ate the pieces of food, and how often their performance lined up with different foraging strategies. A total of 35 chicks between the ages of three to five days were tested, both male and female. Each chick was tested through three different versions of a maze made of 7 X 7 boxes totaling an area of 71.4cm X 71.4cm. These experiments were run over the span of three months, with three to five chicks being tested each week for three different mazes. The findings were then compared to the mathematical calculations of a Machine Learning Model in relation to the Nearest Neighbor Principle and the Optimal Strategy. The results found that the chicks followed the Nearest Neighbor principle in their decisions 68% of the time and the Optimal strategy 62% of the time. The overall goal of this full experiment is to use this biological data to create an accurate AI version of a chick for the Ecotwin virtual ecosystem. This accurate ecosystem model will allow the consequences of human intervention to be analyzed in a safe space first before affecting a real ecosystem to promote ecological conservation.File | Dimensione | Formato | |
---|---|---|---|
final thesis pdfa.pdf
accesso aperto
Dimensione
924.29 kB
Formato
Adobe PDF
|
924.29 kB | Adobe PDF | Visualizza/Apri |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/81092