Další formáty:
BibTeX
LaTeX
RIS
@article{58308,
author = {Pijackova, Kristyna and Nejedlý, Petr and Kremen, Vaclav and Plesinger, Filip and Mivalt, Filip and Lepkova, Kamila and Pail, Martin and Jurak, Pavel and Worrell, Gregory and Brázdil, Milan and Klimes, Petr},
article_location = {BRISTOL},
article_number = {3},
doi = {http://dx.doi.org/10.1088/1741-2552/acdc54},
keywords = {intracranial EEG; genetic algorithms; optimization; neural network; deep learning},
language = {eng},
issn = {1741-2560},
journal = {JOURNAL OF NEURAL ENGINEERING},
title = {Genetic algorithm designed for optimization of neural network architectures for intracranial EEG recordings analysis},
url = {https://iopscience.iop.org/article/10.1088/1741-2552/acdc54},
volume = {20},
year = {2023}
}
TY - JOUR
ID - 58308
AU - Pijackova, Kristyna - Nejedlý, Petr - Kremen, Vaclav - Plesinger, Filip - Mivalt, Filip - Lepkova, Kamila - Pail, Martin - Jurak, Pavel - Worrell, Gregory - Brázdil, Milan - Klimes, Petr
PY - 2023
TI - Genetic algorithm designed for optimization of neural network architectures for intracranial EEG recordings analysis
JF - JOURNAL OF NEURAL ENGINEERING
VL - 20
IS - 3
SP - 1-11
EP - 1-11
PB - IOP PUBLISHING LTD
SN - 1741-2560
KW - intracranial EEG
KW - genetic algorithms
KW - optimization
KW - neural network
KW - deep learning
UR - https://iopscience.iop.org/article/10.1088/1741-2552/acdc54
N2 - Objective. The current practices of designing neural networks rely heavily on subjective judgment and heuristic steps, often dictated by the level of expertise possessed by architecture designers. To alleviate these challenges and streamline the design process, we propose an automatic method, a novel approach to enhance the optimization of neural network architectures for processing intracranial electroencephalogram (iEEG) data. Approach. We present a genetic algorithm, which optimizes neural network architecture and signal pre-processing parameters for iEEG classification. Main results. Our method improved the macro F1 score of the state-of-the-art model in two independent datasets, from St. Anne's University Hospital (Brno, Czech Republic) and Mayo Clinic (Rochester, MN, USA), from 0.9076 to 0.9673 and from 0.9222 to 0.9400 respectively. Significance. By incorporating principles of evolutionary optimization, our approach reduces the reliance on human intuition and empirical guesswork in architecture design, thus promoting more efficient and effective neural network models. The proposed method achieved significantly improved results when compared to the state-of-the-art benchmark model (McNemar's test, p MUCH LESS-THAN 0.01). The results indicate that neural network architectures designed through machine-based optimization outperform those crafted using the subjective heuristic approach of a human expert. Furthermore, we show that well-designed data preprocessing significantly affects the models' performance.
ER -
PIJACKOVA, Kristyna, Petr NEJEDLÝ, Vaclav KREMEN, Filip PLESINGER, Filip MIVALT, Kamila LEPKOVA, Martin PAIL, Pavel JURAK, Gregory WORRELL, Milan BRÁZDIL a Petr KLIMES. Genetic algorithm designed for optimization of neural network architectures for intracranial EEG recordings analysis. \textit{JOURNAL OF NEURAL ENGINEERING}. BRISTOL: IOP PUBLISHING LTD, 2023, roč.~20, č.~3, s.~1-11. ISSN~1741-2560. Dostupné z: https://dx.doi.org/10.1088/1741-2552/acdc54.
|
