Naslov
Advanced Mechanisms of Perception in the Digital Hide and Seek Game Based on Deep Learning

Autori
Livada, Časlav ; Hodak, David

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
2022 International Conference on Smart Systems and Technologies (SST) / - , 2022, 135-140

Skup
International Conference on Smart Systems and Technologies (SST)

Mjesto i datum
Osijek, Hrvatska, 19-20.10.2022

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
agent ; deep learning ; ML-Agents ; perception ; reinforcement learning

Sažetak
The goal of this paper is to demonstrate the power of machine learning and deep learning combined with powerful tools used in creating virtual simulations. This is made possible by the Unity Game Engine, a software used for the development of 2D and 3D video games and simulations, and the MLagents plugin, which contains algorithms for training intelligent agents. The basic principles and algorithms of Deep Learning are described. Agent uses these methods to perceive the environment and solve certain tasks. In this paper, a simulation is created in which the agent tries to find hidden objects using machine learning. When the agent finds an item, a new location for the item is randomly generated. It must also complete the task in a specified amount of time. With the right implementation and parameters, the agent could successfully complete the task.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo

POVEZANOST RADA