Derin evrişimli sinir ağı modellerinin açık kaynak kodlu yazılım platformlarında tasarımının değerlendirilmesi

Öz

Derin evrişimli sinir ağları, iki boyutlu verilerin kullanıldığı, en popüler ve en yaygın derin öğrenme yöntemlerinden birisidir. Özellikle lisans ve lisansüstü öğrencilerin derin öğrenme yöntemlerini özgürce uygulayabilecekleri ve geliştirebilecekleri yeni derin öğrenme modelleri tasarlayabilecekleri, bu konudaki deneyimlerini arttırabilecekleri ortamlara maliyetsiz ve kolayca ulaşabilmeleri, bu gençlerin insanlığa ve bilime hizmet edebilecek bilgi, beceri ve deneyime sahip olmaları açısından çok önemlidir. Açık kaynak kodlu yazılım platformları eğer üniversitelerde ders olarak okutulursa ve öğrencilerin öğrencilik dönemleri boyunca eğitilebilecekleri bir ortama kavuşmaları açısından son derece büyük avantaja sahiptir. Ne var ki günümüzde üniversiteler MATLAB gibi ticari yazılımların lisansını aldıklarında araştırmacıların ulaşabildiği ancak öğrencilerin ulaşamadığı derin öğrenme uygulama ortamları ortaya çıkmaktadır. MATLAB derin öğrenme uygulamalarının gerçekleştirilmesi açısından maliyetli olması dışında önemli bir dezavantajı olmayan bir kapalı kaynak kodlu ticari bir yazılımdır. Bu çalışmada derin evrişimsel sinir ağı modellerinin açık kaynak kodlu yazılım platformlarında tasarımı kaynak araştırması yapılarak ele alınmış ve MATLAB ile kıyaslanmıştır. Açık kaynak kodlu yazılım platformları ile DESA uygulamalarının kolay ulaşılabilir olmasını sağlamak ve gençler arasında popülaritesinin arttırılabilmesi için üniversitelerin müfredat programlarına ders olarak konulması gerekliliği sonucuna varılmıştır.

Anahtar Kelimeler

• Açık kaynak kodlu (AKK),
• Kapalı kaynak kodlu (KKK),
• yazılım,
• MATLAB,
• DESA,
• öğrenme aktarımı

Evaluation of the design of deep convolution neural network models using open source software platforms

Öz

Deep convolutional neural networks (DESA) is one of the most popular and common deep learning method using two-dimensional data. It is especially important for undergraduate and graduate students to have free and easy access to environments where they can freely apply and develop deep learning methods, design new deep learning models, and have knowledge, skills and experience that can serve humanity and science. Open source software platforms have a great advantage if they are taught as a course in universities and in terms of providing an environment where students can be educated during their student period. However, nowadays, when universities obtain the license of commercial software such as MATLAB, deep learning application environments that researchers can reach but cannot reach by students emerge. MATLAB is a commercial software with closed source code that does not have any significant disadvantages other than being costly in terms of realizing deep learning applications. In this study, the design of deep convolutional neural network models on open source software platforms has been handled and compared with MATLAB. It was concluded that open source software should be included in the curriculum of universities in order to make DCNN applications easily accessible and to increase their popularity among young people with open source software platforms.

Kaynakça

1. Ahmed, T., Das, P., Ali, F., Mahmud, F. (2020). A Comparative Study on Convolutional Neural Network Based Face Recognition. 2020 11th International Conference on Computing, Communication and Networking Technologies (ICCCNT), IEEE, 1-3 July, Kharagpur, India.
2. Arora, S., Bhatia, M.P.S. (2018). Handwriting recognition using Deep Learning in Keras. International Conference on Advances in Computing, Communication Control and Networking (ICACCCN2018), IEEE, 12-13 October, Greater Noida (UP), India.
3. Barchi, F., Parisi, E., Urgese, G., Ficarra, E., Acquaviva, A. (2021). Exploration of Convolutional Neural Network models for source code classification. Engineering Applications of Artificial Intelligence, 97, 104075.
4. Bhattacharya, S., Maddikunta., P. K. R., Pham, Q.V., Thippa Reddy Gadekallu., T. R., Krishnan, S. R., Chowdhary, C. L., Alazab, M., Piran, J. (2020). Deep learning and medical image processing for coronavirus (COVID-19) pandemic: A survey. Sustainable Cities and Society, https://doi.org/10.1016/j.scs.2020.102589 , xxx, (xx).
5. Boehmke, B., Hazen, B., Boone, C. A., Robinson, J. L. (2020). A data science and open source software approach to analytics for strategic sourcing, International Journal of Information Management. 54, 102167.
6. Cresson, R. (2019). A Framework for Remote Sensing Images Processing Using Deep Learning Techniques. IEEE Geoscience and Remote Sensing Letters, 16, (1):25-29.
7. Cummings, P. T., Gilmer, J. B. (2019). Open-source molecular modeling software in chemical engineering. Current Opinion in Chemical Engineering, 23:99–105.
8. Dogaru, R., Dogaru, I. (2019). BCONV-ELM: Binary Weights Convolutional Neural Network Simulator based on Keras/Tensorflow, for Low Complexity Implementations. 6th International Symposium on Electrical and Electronics Engineering (ISEEE), 18-20 Oct. Galati, Romania.

9. Duth, S., Raj, S. (2018). Object Recognition in Images using Convolutional Neural Network. Proceedings of the Second International Conference on Inventive Systems and Control (ICISC), 19-20 Jan., Coimbatore, India.
10. Gayathri, S., Varun, P. G., Palanisamy, P. (2020). A lightweight CNN for Diabetic Retinopathy classification from fundus images. Biomedical Signal Processing and Control, 62, 102115.
11. Ghosh, R., Ghosh, K., Maitra, S. (2017). Automatic Detection and Classification of Diabetic Retinopathy stages using CNN. 4th International Conference on Signal Processing and Integrated Networks (SPIN), 2-3 February, Noida, India.
12. Jakhar, K., Hooda, N. (2018). Big Data Deep Learning Framework using Keras: A Case Study of Pneumonia Prediction. 4th International Conference on Computing Communication and Automation (ICCCA) 14-15 Dec. 2018. Greater Noida, India.
13. Jiao, J., Zhao, M., Lin, J., Liang, K. (2020). A comprehensive review on convolutional neural network in machine fault diagnosis. Neurocomputing, (417): 36-63.
14. Jindal, R., Mittal, S.K. (2020). Software reusability metrics estimation for improving stability by clustering base convolution neural network. Materials Today: Proceedings, https://doi.org/10.1016/j.matpr.2020.09.615, xxx, (xx).
15. Kiranyaz, S., Avci, O., Abdeljaber, O., Ince, T., Gabbouj, M., Inman, D.J. (2021). 1D convolutional neural networks and applications: A survey. Mechanical Systems and Signal Processing, 151, 107398.
16. Lin, B.Y., Huang, H.S., Sheu, R. K., Chang, Y.S. (2018). Speech recognition for people with dysphasia using convolutional neural network. IEEE International Conference on Systems, Man, and Cybernetics (SMC), 7-10 Oct. Miyazaki, Japan.
17. Nguyen, P.T., Ruscio, D., Pierantonio, A., Rocco, J.D., Iovino, L. (2021). Convolutional neural networks for enhanced classification mechanisms of metamodels. The Journal of Systems & Software, 172, 110860.
18. Nogay, H. S. (2018). Classification Of Different Cancer Types By Deep Convolutional Neural Networks. Balcan Journal of Electrical&Computer Engineering, 5: 56-59.
19. Nogay, H. S., Akıncı, T. C. (2018). A Convolutional Neural Network Application For Predicting The Locating Of Squamous Cell Carcinoma In The Lung. Balkan Journal of Electrical & Computer Engineering, 6: 207-210.
20. Nogay, H. S. (2017). Deep Convolutional Neural Networks To Detect Lung Cancer Stage”, The Journal of Cognitive Systems, 2: 33-36.
21. Nogay, H. S., Akıncı, T. Ç., Erdemir, G. (2018a). A Convolutional Neural Network Application For The Classification Of Lung Cancer Types. Academic Journal Industrial Technologies, 5:7-12,.
22. Nogay, H. S., Akıncı, T. Ç., Erdemir, G. (2018b) “Estimation Of Head & Neck Cancer Stage By Using Deep Convolutional Neural Networks. Academic Journal Industrial Technologies, 5: 13-19.
23. Nogay, H. S., Adeli, H. (2020). Machine learning (ML) for the diagnosis of autism spectrum disorder (ASD) using brain imaging, Reviews In The Neurosciences, DOI: 10.1515/revneuro-2020-0043, 1-17.
24. Nogay, H. S., Akıncı T. Ç. (2020). Classification of operation cases in electric arc welding wachine by using deep convolutional neural networks, Neural Computing & Applications, https://doi.org/10.1007/s00521-020-05436-y, 1-14.
25. Rai, P., Londhe, N.D., Raj, R. (2020). Fault classification in power system distribution network integrated with distributed generators using CNN. Electric Power Systems Research, https://doi.org/10.1016/j.epsr.2020.106914, xxx(xx):xx.
26. Rodrigues, C. A. S. P., Vinhal, C., Cruz, G. (2017). Fully convolutional networks for segmenting images from an embedded camera. 2017 IEEE Latin American Conference on Computational Intelligence (LA-CCI), IEEE, 8-10 November, Arequipa, Peru.
27. Qiao, S., Ma, J. (2018). A Face Recognition System Based on Convolution Neural Network. Chinese Automation Congress (CAC), IEEE, 30 Nov.-2 Dec. Xi'an, China.
28. Qin, J., Pan, W., Xiang, X., Tan, Y., Hou, G. (2020). A biological image classification method based on improved CNN. Ecological Informatics, 58, 101093.
29. Quiñonez, Y., Carmen Lizarraga, C., Peraza, J., NZatarain, O. (2020). Image recognition in UAV videos using convolutional neural networks. The Institution of Engineering and Technology IET Software, 14 (2):176-181.
30. Walt, S. (2003). Free/Open Source Software Development Practices in the Computer Game Community. University of California- Institute for Software Research, http://www.ics.uci.edu/~redmiles/ics221-FQ03/paper.
31. Yang, Y., Nie Z., Huang, S., Lin, P., Wu, J. (2019). Multilevel Features Convolutional Neural Network for Multifocus Image Fusion. IEEE Transactions on Computational Imaging, 5(2): 262-273.
32. Yuan, L., Qu, Z., Zhao, Y., Zhang, H., Nian, Q. (2017). A Convolutional Neural Network based on TensorFlow for Face Recognition. 2017 IEEE 2nd Advanced Information Technology, Electronic and Automation Control Conference (IAEAC), 25-26 March, Chongqing, China.