TY - GEN
T1 - An auto contrast custom convolutional neural network to identifying Gram-negative bacteria
AU - Satoto, Budi Dwi
AU - Utoyo, Mohammad Imam
AU - Rulaningtyas, Riries
AU - Koendhori, Eko Budi
N1 - Funding Information:
ACKNOWLEDGMENT The author would like to thank Dr. Soetomo Surabaya hospital's microbiology laboratory for providing the primary data of pneumonia patients. The Ministry of Research and Higher Education supports this work, Grant 2020, with contract number: 812 / UN3.14 / PT / 2020.
Publisher Copyright:
© 2020 IEEE.
PY - 2020/11/17
Y1 - 2020/11/17
N2 - The process of identifying bacteria is an essential factor in the medical field. One of the germs that cause lung damage in pneumonia is gram-negative bacteria. The convolutional neural network method is the newest approach to machine learning because it has a high degree of accuracy. But the drawback is due to his in-depth knowledge, the computation time for the training process takes a long time. The method offered in this research is automatic contrast addition in the preprocessing stage and the use of custom layers. Also, augmentation data added to increase the variation in the amount of data in the training process. In using custom layers, the objective is to obtain minimal computational training time while maintaining maximum accuracy values. The results show that an average accuracy around 98.59% with average training time around 01 minutes 56 seconds, average MSE 0.0274, RMSE 0.1693, and MAE 0.0185.
AB - The process of identifying bacteria is an essential factor in the medical field. One of the germs that cause lung damage in pneumonia is gram-negative bacteria. The convolutional neural network method is the newest approach to machine learning because it has a high degree of accuracy. But the drawback is due to his in-depth knowledge, the computation time for the training process takes a long time. The method offered in this research is automatic contrast addition in the preprocessing stage and the use of custom layers. Also, augmentation data added to increase the variation in the amount of data in the training process. In using custom layers, the objective is to obtain minimal computational training time while maintaining maximum accuracy values. The results show that an average accuracy around 98.59% with average training time around 01 minutes 56 seconds, average MSE 0.0274, RMSE 0.1693, and MAE 0.0185.
KW - Gram-negative bacteria
KW - Pneumonia
KW - auto contrast Convolutional neural network
KW - data augmentation
UR - http://www.scopus.com/inward/record.url?scp=85099642757&partnerID=8YFLogxK
U2 - 10.1109/CENIM51130.2020.9297964
DO - 10.1109/CENIM51130.2020.9297964
M3 - Conference contribution
AN - SCOPUS:85099642757
T3 - CENIM 2020 - Proceeding: International Conference on Computer Engineering, Network, and Intelligent Multimedia 2020
SP - 70
EP - 75
BT - CENIM 2020 - Proceeding
PB - Institute of Electrical and Electronics Engineers Inc.
T2 - 2020 International Conference on Computer Engineering, Network, and Intelligent Multimedia, CENIM 2020
Y2 - 17 November 2020 through 18 November 2020
ER -