EUVIMEDDécouvrir EUVIMED
Open access · CC BY 4.0

Effect of hyper-parameters on the performance of ConvLSTM based deep neural network in crop classification

Awab Ur Rashid Durrani iD, Nasru Minallah, Najam Aziz, Jaroslav Frnda, Waleed Khan, Jan Nedoma iD

PLOS ONE · 2023-02-09 · 18 (2) · e0275653

DOI: 10.1371/journal.pone.0275653 ↗

At a glance

Journal article in PLOS ONE, published online on 2023-02-09. Written by Awab Ur Rashid Durrani and 5 co-authors. Institution: Department of Computer Systems Engineering, University of Engineering and Technology Peshawar (UETP), Peshawar, Pakistan. Major MEDLINE topics: Neural Networks, Computer. Cited 17 times so far (Crossref); 40 references. The full text is freely available (open access, CC BY 4.0). Funded by: VSB-Technical University of Ostrava.

RésuméSubjectsRéférences bibliographiquesCiteRelated studies

Abstract (original language)

Deep learning based data driven methods with multi-sensors spectro-temporal data are widely used for pattern identification and land-cover classification in remote sensing domain. However, adjusting the right tuning for the deep learning models is extremely important as different parameter setting can alter the performance of the model. In our research work, we have evaluated the performance of Convolutional Long Short-Term Memory (ConvLSTM) and deep learning techniques, over various hyper-parameters setting for an imbalanced dataset and the one with highest performance is utilized for land-cover classification. The parameters that are considered for experimentation are; Batch size, Number of Layers in ConvLSTM model, and No of filters in each layer of the ConvLSTM are the parameters that will be considered for our experimentation. Experiments also have been conducted on LSTM model for comparison using the same hyper-parameters. It has been found that the two layered ConvLSTM model having 16-filters and a batch size of 128 outperforms other setting scenarios, with an overall validation accuracy of 97.71%. The accuracy achieved for the LSTM is 93.9% for training and 92.7% for testing.

Subjects

MeSH terms (MEDLINE)

Neural Networks, Computer

Authors and institutions

  • Department of Computer Systems Engineering, University of Engineering and Technology Peshawar (UETP), Peshawar, Pakistan.
  • National Center for Big Data and Cloud Computing (NCBC), University of Engineering and Technology Peshawar (UETP), Peshawar, Pakistan.
  • Department of Quantitative Methods and Economic Informatics, Faculty of Operation and Economics of Transport and Communication, University of Zilina, Zilina, Slovakia.
  • Department of Telecommunications, Faculty of Electrical Engineering and Computer Science, VSB - Technical University, Ostrava-Poruba, Czechia of Ostrava, Ostrava-Poruba, Czechia.

ORCID profiles: Awab Ur Rashid Durrani, Jan Nedoma

Funding

  • VSB-Technical University of Ostrava

Cite

APA

Durrani, A. U. R., Minallah, N., Aziz, N., Frnda, J., Khan, W., & Nedoma, J. (2023). Effect of hyper-parameters on the performance of ConvLSTM based deep neural network in crop classification. PLOS ONE, 18(2), e0275653. https://doi.org/10.1371/journal.pone.0275653

Vancouver

Durrani AUR, Minallah N, Aziz N, Frnda J, Khan W, Nedoma J. Effect of hyper-parameters on the performance of ConvLSTM based deep neural network in crop classification. PLOS ONE. 2023;18(2):e0275653. doi:10.1371/journal.pone.0275653

BibTeX
@article{Durrani2023,
  author = {Durrani, Awab Ur Rashid and Minallah, Nasru and Aziz, Najam and Frnda, Jaroslav and Khan, Waleed and Nedoma, Jan},
  title = {Effect of hyper-parameters on the performance of ConvLSTM based deep neural network in crop classification},
  journal = {PLOS ONE},
  year = {2023},
  volume = {18},
  number = {2},
  pages = {e0275653},
  doi = {10.1371/journal.pone.0275653}
}
Start for free →