20200303_Algal Morphological Identification in Watersheds for Drinking Water Supply Using Neural Architecture Search for Convolutional Neural Network
1. Title, Journal, and Authors
Title: Algal Morphological Identification in Watersheds for Drinking Water Supply Using Neural Architecture Search for Convolutional Neural Network
Journal: Water
Authors: Jungsu Park1, Hyunho Lee2, Cheol Young Park3, Samiul Hassan4, Tae-Young Heo5, and Woo Hyoung Lee4,*
1 Water Quality & Safety Research Center, Korea Water Resources Corporation, 200 Sintanjin-Ro, Daedeok-Gu, Daejeon 34350, Korea
2 Water Data Collection and Analysis Department, Korea Water Resources Corporation, 200 Sintanjin-Ro, Daedeok-Gu, Daejeon 34350, Korea
3 The C41 & Cyber Center, George Mason University, MS 4B4 Fairfax, VA 22030, USA
4 Department of Civil, Environmental and Construction Engineering, University of Central Florida, 12800 Pegasus Dr.Suite 211, Orlando, FL 32816-2450, USA
5 Department of Information & Statistics, Chhungbuk National University, Chungdae-Ro 1, SewWon-Gu, Cheongju, Chungbuk 28644, Korea
* Correspondence: woohyung.lee@ucf.edu; Tel.: +1-407-823-5304
2. Summary
A CNN model was carried to classify the 8 algal morphological cell images based on NAS which finds out the parameters for the AI analysis. To compare the model’s performances, an ordinary CNN model, a CNN model based on ReNet, and a CNN model based on NAS were introduced.
In the results, the model based on NAS showed the highest accuracy, and when using augmented data had higher accuracy than using raw data. Because only 8 algae genera and their cell images were introduced, this point would be needed to improve, afterward.
3. Originality & Creativity
Maybe it would be the creative paper to introduce the NAS method and classify the algae cell images.
4. Contact
Dae Seong Jeong / Integrated Ph.D. program
Environmental Systems Engineering Lab.
School of Environmental Science & Engineering
Gwangju Institute of Science and Technology
1 Oryong-dong Buk-gu Gwangju, 500-712, Korea
Phone : +82-10-2003-7860
E-mail : jeongds92@gist.ac.kr