Deep Learning in Modeling Complex Systems

No Thumbnail Available

Files

URL

Journal Title

Journal ISSN

Volume Title

Perustieteiden korkeakoulu | Bachelor's thesis
Electronic archive copy is available locally at the Harald Herlin Learning Centre. The staff of Aalto University has access to the electronic bachelor's theses by logging into Aaltodoc with their personal Aalto user ID. Read more about the availability of the bachelor's theses.
Unless otherwise stated, all rights belong to the author. You may download, display and print this publication for Your own personal use. Commercial use is prohibited.

Authors

Date

2024-12-13

Department

Major/Subject

Data Science

Mcode

SCI3095

Degree programme

Aalto Bachelor’s Programme in Science and Technology

Language

en

Pages

23+3

Series

Abstract

On many occasions, there arises a need to model complex system behavior to enable the management and/or inference of given systems. Common examples include managing resource efficiency and simulating security threats whose analysis is prone to immense complexity. For such tasks, traditional human-made mathematical models have been applied in the past, but these traditional approaches are becoming increasingly challenging to implement as system complexity grows. To overcome the rise in complexity, the application of deep learning methods has emerged as a solution. The aim of this thesis is to review applicable deep learning methods for complex systems modeling. After a review of the literature, there is evidence of an abundance of successful deep learning applications across a wide spectrum of fields. In particular, graph neural networks (GNNs) and attention mechanisms show high potential in weather prediction and other dynamic systems simulations. The ability of these methods to represent high-dimensional and non-linear input spaces justifies their application in complex systems scenarios. Even with these strengths, most GNN-based systems are still proofs of concept and highly tuned for single applications. Thus, this review points to the development of robust field- and problem-specific frameworks to overcome the uncertainty in method selection present in the current scene. For this task, the thesis provides a starting point of relevant research with methods applicable for the advancement of these frameworks.

Description

Supervisor

Korpi-Lagg, Maarit

Thesis advisor

Hirvisalo, Vesa

Keywords

deep learning, complex system, graph neural network, robustness, systems modeling

Other note

Citation

Permanent link to this item

https://urn.fi/URN:NBN:fi:aalto-202502182436

Collections

[kand] Perustieteiden korkeakoulu / SCI