Univerza na Primorskem Fakulteta za matematiko, naravoslovje in informacijske tehnologije
SI | EN

Ponedeljkov seminar računalništva in informatike - Arhiv

2025 2024 2023 2022 2021 2020 2019 2018 2017 2016 2015 2014 2013 2012
1 2 3 4 5 6 7 8 9 10 11 12

ponedeljek, 19. maj 2025 Vanja MILESKI: Deep learning for time series data using Inception and ResNet paradigms

V ponedeljek, 19. maja 2025, bo ob 16:00 uri izvedeno
predavanje v okviru PONEDELJKOVEGA SEMINARJA RAČUNALNIŠTVA IN INFORMATIKE
Oddelkov za Informacijske znanosti in tehnologije UP FAMNIT in UP IAM.

ČAS/PROSTOR: 19. maj 2025 ob 16.00 v FAMNIT-VP2.

----------------------------------------
PREDAVATELJ: Vanja MILESKI
----------------------------------------

After graduating from the Faculty of Computer and Information Science at the University of Ljubljana in 2015, Vanja Mileski started working at the Jožef Stefan Institute (JSI). He was a Master's student at the International Postgraduate School Jožef Stefan and a student researcher at the JSI. After finishing his Master's studies, he applied his knowledge of data mining in the private sector as a Data Scientist in the retail, telecommunications, banking, stock market and insurance sectors. His current research interests include Time-Series classification, Deep Learning, ResNet and Inception architectures as well as LLMs.

-----------------------------------------------------------------------------------------------------------------
NASLOV: Deep learning for time series data using Inception and ResNet paradigms
-----------------------------------------------------------------------------------------------------------------

POVZETEK:

Prediction of customer behaviour represents a pivotal task in the retail sector, focusing on identifying customers with a high risk of attrition. We analyse a multi-year dataset from a large Slovenian retailer, encompassing detailed customer demographics, purchasing behaviours, and spending habits, represented as time series. We incorporate Convolutional Neural Networks (CNNs) and apply them on the time series, where the features can be viewed as one-dimensional vectors. Furthermore, we incorporate ResNet-like skip connections, as well as Inception modules with different kernel sizes in our architecture to improve the predictive performance compared to traditional ML techniques.

Seminar bo potekal v živo, s pričetkom ob 16:00 uri v učilnici FAMNIT-VP2.

Vabljeni.