Primjena strojnog učenja u kemijskom inženjerstvu

undergraduate thesis

• Title (english): Application of machine learning in chemical engineering
• Author: Nela Rapinac
• Mentor: Miroslav Jerković (mentor)
• Committee member: Miroslav Jerković (predsjednik povjerenstva)
• Committee member: Erna Begović Kovač (član povjerenstva)
• Committee member: Željka Ujević Andrijić (član povjerenstva)
• Granter: University of Zagreb; Faculty of Chemical Engineering and Technology; Zagreb
• Defense date and country: 2023-09-19, Croatia
• Scientific / art field, discipline and subdiscipline: TECHNICAL SCIENCES; Computing

Abstract

Strojno učenje je jedan od najnovijih i trenutno najpopularnijih dijelova računalne znanosti - primjena algoritama koji stvaraju neki smisao od skupa podataka. Kreće se od vektora, a glavni već razvijeni algoritmi su regresija i klasifikacija. Od složenijih algoritama najpopularnije su neuralne mreže koje, kao što im i ime kaže, podsjećaju na rad ljudskog mozga te se sastoje od neurona i čvorova. Strojno učenje, uključujući i neuralne mreže, ima široku primjenu i postaje svakodnevnica uporabe, a da se to ni ne primjećuje. Strojno učenje se može primjenjivati i u kemijskom inženjerstvu, a posebno je popularno razvijanje metoda za predviđanje molekularnih svojstava. Postoje razne metode i modeli, a jedna od metoda je i Uni-Mol metoda koja uzima u obzir 3D konfiguraciju molekula. Metoda se bazira na transformeru, vrsti dubokog učenja koje radi slično kao i neuralne mreže uz određena poboljšanja. U ovome radu bit će objašnjeni spomenuti pojmovi i ključne riječi te algoritmi i metode, a cilj rada je uvidjeti veliki potencijal koje strojno učenje ima u konkretno kemijskom inženjerstvu, ali i općenito u ostalim industrijama.

Abstract (english)

Machine learning is one of the newest and currently most popular parts of computer science - the application of algorithms that make some sense out of a set of data. It starts from vectors, and the main already developed algorithms are regression and classification. Among the more complex algorithms, the most popular are neural networks, which, as their name suggests, resemble the work of the human brain and consist of neurons and nodes. Machine learning, including neural networks, have a wide application and becoming common place without anyone realizing it. Machine learning can also be applied in chemical engineering, and the development of methods for predicting molecular properties is particularly popular. There are various methods and models, and one of the methods is the Uni-Mol method, which takes into account the 3D configuration of molecules. The method is based on a transformer, a type of deep learning that works similarly to neural networks with some improvements. In this paper, the mentioned terms and keywords as well as algorithms and methods will be explained, and the aim of the paper is to comprehend the great potential that machine learning has in chemical engineering in particular, but also in other industries in general.

• Keywords: strojno učenje; vektor; neuralne mreže; transformer; predviđanje svojstva molekula; Uni-Mol metoda; Python
• Keywords (english): machine learning; vector; neural networks; transformer; molecular property prediction; Uni-Mol method; Python
• Language: croatian
• URN:NBN: urn:nbn:hr:149:794249
• Study programme: Title: Chemical Engineering - Undergraduate study; Study programme type: university; Study level: undergraduate; Academic / professional title: sveučilišni/a prvostupnik/prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka kemijskog inženjerstva (sveučilišni/a prvostupnik/prvostupnica (baccalaureus/baccalaurea) inženjer/inženjerka kemijskog inženjerstva)
• Type of resource: Text
• File origin: Born digital
• Access conditions: Open access
• Created on: 2023-11-28 10:40:32