Project detail

Artificial neural networks weight initialization for classification of spectroscopic data

Duration: 01.03.2021 — 28.02.2022

Funding resources

Brno University of Technology - Vnitřní projekty VUT

- whole funder (2021-01-01 - 2022-12-31)

On the project

The project addresses the problem of automatized classification of spectroscopic data by artificial neural networks. The biggest challenges in this field are a huge over-parametrization (and computational cost), interpretability and model generalizability. The proposed solution is based on a novel strategy for weight initialization of the network, using spectra modeling. This results in a rapid reduction of parameters and training time, with slightly better performance and interpretability.

Mark

CEITEC-J-21-7542

Default language

Czech

People responsible

Kaiser Jozef, prof. Ing., Ph.D. - fellow researcher
Vrábel Jakub, Ing. - principal person responsible

Units

Advanced instrumentation and methods for material characterization
- internal (2021-01-01 - 2021-12-31)
Central European Institute of Technology BUT
- beneficiary (2021-01-01 - 2021-12-31)

Results

VRÁBEL, J.; KÉPEŠ, E.; POŘÍZKA, P.; KAISER, J. Artificial neural network weights penalization and initialization for spectroscopic data. 2021.
Detail

Responsibility: Vrábel Jakub, Ing.