izpis_h1_title_alt

Analiza vpliva parametrov modela na evolucijo skupinskega vedenja
ID Mišanović, Danijel (Author), ID Lebar Bajec, Iztok (Mentor) More about this mentor... This link opens in a new window

.pdfPDF - Presentation file, Download (7,81 MB)
MD5: 5FA56A34F8A2D6746DDEFEFCE7EEB908

Abstract
V naravi lahko srečamo veliko različnih oblik skupinskega vedenja kot so jate ptic ali rib, črede pašnih živali in roji insektov. Ker je natančno modeliranje vedenja živali in njihovega naravnega okolja zelo kompleksno, si pri raziskovanju evolucije skupinskega vedenja pomagamo s poenostavljenimi modeli živali in okolja. Iz njih želimo izluščiti zakonitosti, pravila in znanja zakaj in kako do takega vedenja pride. Pri teh modelih raziskovalci največkrat opazujejo lastnosti in značilnosti, kot so na primer, kakšen je prenos informacije med entitetami, kako združevanje v skupino vpliva na obrambo pred plenilci in na hranjenje, kakšni režimi vedenja se razvijejo ter kakšen je proces odločanja skupine glede na različne situacije (oblike habitata, razmerje različnih plenilcev). V naši magistrski nalogi smo se odločili razširiti model in analizo, ki so jo opravili Demšar in sodelavci [doi: 10.1371/journal.pone.0168876, 10.1038/srep39428]. Pri tem smo se osredotočili predvsem na to kako oblika habitata, razmerje različnih plenilcev, lokacija pojavitve plenilca in tip simulacijske zanke, vpliva na razvoj štirih različnih režimov vedenja skupine: močno usklajeno gibanje, kroženje okoli praznega jedra, neusklajeno prepletanje in režim tranzicije. Za oblike habitata smo poleg že implementiranih kroga in kvadrata, implementirali še obliko neskončne mreže in poligona. Pri lokaciji pojavitve plenilca smo poleg pojavitve izven habitata, analizirali še vpliv pojavitve plenilca znotraj habitata. Na koncu smo še primerjali vpliv različnih implementacij asinhrone simulacijske zanke v nasprotju s sinhrono simulacijsko zanko. V primeru različnih simulacijskih zank nas je zanimalo predvsem, ali tudi na našem modelu držijo zadnje ugotovitve na tem področju, da asinhronost v simulacijski zanki povečuje verjetnost za pojav močno usklajenega gibanja.

Language:Slovenian
Keywords:genetski algoritmi, mehka logika, evolucija, simulacija, skupinsko vedenje, odziv na plenilca, taktike napada, vpliv habitata
Work type:Master's thesis/paper
Organization:FRI - Faculty of Computer and Information Science
Year:2019
PID:20.500.12556/RUL-112902 This link opens in a new window
COBISS.SI-ID:1538468035 This link opens in a new window
Publication date in RUL:20.11.2019
Views:1577
Downloads:253
Metadata:XML RDF-CHPDL DC-XML DC-RDF
:
Copy citation
Share:Bookmark and Share

Secondary language

Language:English
Title:Senistivity analysis of the evolution of group behaviour model
Abstract:
In nature we can find various shapes of collective behavior such as flocks of birds, schools of fish, swarms of insects and herds of grazing animals. In research of evolution of group behavior accurate modelling of animal behavior and their natural habitat is to complex therefor we help ourselves with simplified models of animals and their habitats. We try to find different rules, properties and knowledge as to why and how collective behavior came to evolve. In these models, researchers mostly investigate various properties of collective behavior like: transfer of information between entities, benefits of grouping (defense against predators and foraging), group behavior types and group decision-making processes based on different situations (different habitats or ratio of different predators). In our master thesis we expanded the model and analysis of Demšar and co-workers [doi: 10.1371/journal.pone.0168876, 10.1038/srep39428]. We were mostly focused on how different shapes of habitats, ratio of different predators, spawn locations of predators and type of simulation loop influence development of four different types of group behavior regimes: polarized, milling, swarming and transition. We implemented two additional habitats, infinite lattice and polygon to add to the set of already implemented circle and square. We also analyzed how spawning predators within living areas differentiates from spawning predators outside of a living area. At the end we also compared the effects of different implementations of asynchronous simulation loop in comparison to synchronized simulation loop. We were especially interested if the latest findings in this field, that asynchronous simulation loop increases the chance for strongly polarized regime to evolve is also true for our model.

Keywords:genetic algorithms, fuzzy logic, evolution, simulation, collective behavior, reaction to predator, attack tactics, influence of habitat

Similar documents

Similar works from RUL:
Similar works from other Slovenian collections:

Back