Determining critical factors for the success of machine learning libraries considering fuzzy interrelationships

Öz

Advances in Machine Learning have led to groundbreaking developments across various fields. A key driver of this progress is the growing capabilities of machine learning libraries. However, these libraries still face certain limitations, which can hinder their application in some areas. In particular, developers often need to integrate their solutions across different libraries, languages, and platforms, which can reduce the effectiveness of machine learning applications. This study examines the direct and indirect impacts of various factors on the success of machine learning libraries, leveraging expert evaluations. Focusing on critical factors that shape library development based on user experiences is crucial. This approach ensures strategic advancement and optimizes energy and resource use. Given that many of these factors involve human perception-related uncertainties, the methodology incorporates a fuzzy linguistic approach with z-numbers to address the inherent vagueness. The findings reveal that considering the interrelationships among factors significantly alters their importance, as it accounts for both direct and indirect influences. Moreover, these interrelationships highlight which factors should be strategically prioritized, as improvements in one factor can drive improvements in others. The results obtained can support guiding the community developing artificial intelligence libraries toward the right objectives and formulating effective strategies to ensure the sustainability of artificial intelligence transformation.

Anahtar Kelimeler

• Artificial intelligence
• Fuzzy logic
• Linguistic term set
• Machine learning libraries
• Z-number

Makine öğrenmesi kütüphanelerinin başarısı için kritik faktörlerin bulanık karşılıklı ilişkiler göz önünde bulundurularak belirlenmesi

Öz

Makine öğrenmesindeki ilerlemeler, çeşitli alanlarda çığır açan gelişmelere yol açmıştır. Bu ilerlemenin temel itici güçlerinden biri, makine öğrenmesi kütüphanelerinin artan yetenekleridir. Ancak, bu kütüphaneler hâlâ bazı sınırlamalara sahiptir ve bu durum, bazı alanlardaki uygulamalarını engelleyebilmektedir. Özellikle geliştiriciler, çözümlerini farklı kütüphaneler, diller ve platformlar arasında entegre etmek zorunda kaldıklarında, makine öğrenmesi uygulamalarının etkinliği azalabilmektedir. Bu çalışma, uzman değerlendirmelerine dayanarak, çeşitli faktörlerin makine öğrenmesi kütüphanelerinin başarısı üzerindeki doğrudan ve dolaylı etkilerini incelemektedir. Zira kullanıcı deneyimlerine dayalı olarak kütüphane geliştirimini şekillendiren kritik faktörlere odaklanmak, stratejik ilerleme sağlamak ve enerji ile kaynak kullanımını optimize etmek açısından büyük önem taşımaktadır. Bu faktörlerin çoğu insan algısına bağlı belirsizlikler içerdiğinden, yöntem olarak z-sayılarla birlikte bulanık dilsel bir yaklaşım benimsenmiştir. Bulgular, faktörler arasındaki karşılıklı ilişkilerin dikkate alınmasının, onların önem derecelerini önemli ölçüde değiştirdiğini ortaya koymuştur; zira bu yaklaşım hem doğrudan hem de dolaylı etkileri hesaba katmaktadır. Ayrıca bu ilişkiler, hangi faktörlerin stratejik olarak önceliklendirilmesi gerektiğini de vurgulamaktadır; çünkü bir faktördeki iyileşme, diğerlerinde de gelişmeleri tetikleyebilir. Elde edilen sonuçlar, yapay zekâ kütüphanelerini geliştiren topluluğun doğru hedeflere yönelmesine ve yapay zekâ dönüşümünün sürdürülebilirliğini sağlamak için etkili stratejiler formüle etmesine katkı sağlayabilir.

Anahtar Kelimeler

• Yapay zekâ
• Bulanık mantık
• Dilsel terim kümesi
• Makine öğrenmesi kütüphaneleri
• Z-sayısı

Kaynakça

1. [1] Wang Z, Liu K, Li J, Zhu Y, Zhang Y. “Various frameworks and libraries of machine learning and deep learning: a survey”. Archives of Computational Methods in Engineering, 1-24, 2019.
2. [2] Al-Bdour G, Al-Qurran R, Al-Ayyoub M, Shatnawi A. “A detailed comparative study of open source deep learning frameworks”. ArXiv Preprint, arXiv:1903.00102, 2019.
3. [3] Han J, Deng S, Lo D, Zhi C, Yin J, Xia X. “An empirical study of the dependency networks of deep learning libraries”. 2020 IEEE International Conference on Software Maintenance and Evolution (ICSME), 2020.
4. [4] Wei A, Deng Y, Yang C, Zhang L. “Free lunch for testing: Fuzzing deep-learning libraries from open source”. Proceedings of the 44th International Conference on Software Engineering, 2022.
5. [5] Bahrampour S, Ramakrishnan N, Schott L, Shah M. “Comparative study of deep learning software frameworks”. ArXiv Preprint, arXiv:1511.06435, 2015.
6. [6] Nejadgholi M, Yang J. “A study of oracle approximations in testing deep learning libraries”. 2019 34th IEEE/ACM International Conference on Automated Software Engineering (ASE), 2019.
7. [7] Wang S, Shrestha N, Subburaman AK, Wang J, Wei M, Nagappan N. “Automatic unit test generation for machine learning libraries: How far are we?”. 2021 IEEE/ACM 43rd International Conference on Software Engineering (ICSE), 2021.
8. [8] Nguyen G, Dlugolinsky S, Bobák M, Tran V, García ÁL, Heredia I, Malík P, Hluchý L. “Machine learning and deep learning frameworks and libraries for large-scale data mining: a survey”. Artificial Intelligence Review, 52, 77-124, 2019.

9. [9] Dai W, Berleant D. “Benchmarking contemporary deep learning hardware and frameworks: A survey of qualitative metrics”. 2019 IEEE First International Conference on Cognitive Machine Intelligence (CogMI), 2019.
10. [10] Shatnawi A, Al-Bdour G, Al-Qurran R, Al-Ayyoub M. “A comparative study of open source deep learning frameworks”. 2018 9th international conference on information and communication systems (icics), 2018.
11. [11] Al-Bdour G, Al-Qurran R, Al-Ayyoub M, Shatnawi A. “Benchmarking open source deep learning frameworks”. International Journal of Electrical & Computer Engineering, 10(5), 5479-5486, 2020.
12. [12] Gevorkyan MN, Demidova AV, Demidova TS, Sobolev AA. “Review and comparative analysis of machine learning libraries for machine learning”. Discrete and Continuous Models and Applied Computational Science, 27(4), 305-315, 2019.
13. [13] Erickson BJ, Korfiatis P, Akkus Z, Kline T, Philbrick K. “Toolkits and libraries for deep learning”. Journal of Digital Imaging, 30, 400-405, 2017.
14. [14] Gheisari M, Ebrahimzadeh F, Rahimi M, Moazzamigodarzi M, Liu Y, Pramanik PKD, Heravi MA, Mehbodniya A, Ghaderzadeh M, Feylizadeh MR, Kosari S. “Deep learning: Applications, architectures, models, tools, and frameworks: A comprehensive survey”. CAAI Transactions on Intelligence Technology, 8, 581–606, 2023.
15. [15] Alawi ZB. “A comparative survey of PyTorch vs TensorFlow for deep learning: Usability, performance, and deployment trade-offs”. ArXiv Preprint, arXiv:2508.04035, 2025.
16. [16] Bafghi RA, Raissi M. “Comparing PINNs across frameworks: JAX, tensorflow, and pytorch”. ICLR 2024 Workshop on AI4DifferentialEquations in Science, 2024.
17. [17] Bećirović M, Kurtović A, Smajlović N, Kapo M, Akagić A. “Performance comparison of medical image classification systems using TensorFlow Keras, PyTorch, and JAX”. ArXiv Preprint, arXiv:2507.14587, 2025.
18. [18] Tchakoute RN, Tadonki C, Dokladal P, Mesri Y. “Benchmark-based Study of CPU/GPU Power-Related Features through JAX and TensorFlow”. ArXiv Preprint, arXiv:2505.03398, 2025.
19. [19] Ratul IJ, Zhou Y, Yang K. “Accelerating deep learning inference: A comparative analysis of modern acceleration frameworks”. Electronics, 14(15), 2977, 2025.
20. [20] Jovic A, Brkic K, Bogunovic N. “An overview of free software tools for general data mining”. 37th international convention on information and communication technology, electronics and microelectronics (MIPRO), Opatija, Croatia, 2014.
21. [21] Rajagopalan G. “Data visualization with python libraries”. Editors: Rajagopalan G. A Python Data Analyst’s Toolkit: Learn Python and Python-based Libraries with Applications in Data Analysis and Statistics, 243-278, 2021.
22. [22] Sial AH, Rashdi SYS, Khan AH. “Comparative analysis of data visualization libraries Matplotlib and Seaborn in Python”. International Journal of Advanced Trends in Computer Science and Engineering, 1(10), 277-281, Ocak 2021.
23. [23] Rubashkin M. “Getting started with deep learning”. 2017. http://www.kdnuggets.com/2017/03/getting-started-deep-learning.html (25.10.2022).
24. [24] Google Open Source. “TensorFlow”. n.d.. https://opensource.google/projects/tensorflow (15.09.2023).
25. [25] Meta Open Source. “PyTorch”. n.d.. https://opensource.fb.com/projects/pytorch/ (09.10.2023).
26. [26] Shi S, Wang Q, Xu P, Chu X. “Benchmarking state-of-the-art deep learning software tools”. 7th International Conference on Cloud Computing and Big Data (CCBD), 2016.
27. [27] Preston-Werner T, Wanstrath C, Hyett PJ, Chacon S. “GitHub”. 8 Şubat 2008. https://github.com/ (25.09.2023).
28. [28] Zhou L, Pan S, Wang J, Vasilakos AV. “Machine learning on big data: Opportunities and challenges”. Neurocomputing, 350-361, 10 Mayıs 2017.
29. [29] Kim T, Zhou X, Pendyala RM. “Computational graph-based framework for integrating econometric models and machine learning algorithms in emerging data-driven analytical environments”. Transportmetrica A: Transport Science, 3(18), 1346-1375, 2 Aralık 2022.
30. [30] ProjectPro. “15 popular machine learning frameworks for model training”. 2023. https://www.projectpro.io/article/machine-learning-frameworks/509 (12.08.2023).
31. [31] May M. “An overview of python deep learning frameworks”. 2017. https://www.kdnuggets.com/2017/02/python-deep-learning-frameworks-overview.html (15.05.2023).
32. [32] Hegde V, Usmani S. Parallel and distributed deep learning. Stanford University, 2016.
33. [33] Nutt PC. “Comparing methods for weighting decision criteria”. Omega, 8(2), 163-172, 1980.
34. [34] Odu GO. “Weighting methods for multi-criteria decision making technique”. Journal of Applied Sciences and Environmental Management, 23(8), 1449-1457, 2019.
35. [35] Roszkowska E. “Rank ordering criteria weighting methods–a comparative overview”. Optimum. Studia Ekonomiczne, 5(65), 14-33, 2013.
36. [36] De Feo G, De Gisi S. “Using an innovative criteria weighting tool for stakeholders involvement to rank MSW facility sites with the AHP”. Waste Management, 30(11), 2370-2382, 2010.
37. [37] Mardani A, Nilashi M, Zakuan N, Loganathan N, Soheilirad S, Saman MZM, Ibrahim O. “A systematic review and meta-Analysis of SWARA and WASPAS methods: Theory and applications with recent fuzzy developments”. Applied Soft Computing, 57, 265-292, 2017.
38. [38] Saeidi P, Mardani A, Mishra AR, Cajas VEC, Carvajal MG. “Evaluate sustainable human resource management in the manufacturing companies using an extended Pythagorean fuzzy SWARA-TOPSIS method”. Journal of Cleaner Production, 370, 133380, 2022.
39. [39] İlbahar E, Çolak M, Karaşan A, Kaya İ. “A combined methodology based on Z-fuzzy numbers for sustainability assessment of hydrogen energy storage systems”. International Journal of Hydrogen Energy, 47(34), 15528-15546, 2022.
40. [40] Karaşan A, Kaya İ, Erdoğan M, Çolak M. “A multicriteria decision making methodology based on two-dimensional uncertainty by hesitant Z-fuzzy linguistic terms with an application for blockchain risk evaluation”. Applied Soft Computing, 113, 108014, 2021.
41. [41] Kaya İ, Işık G, Karaşan A, Kutlu Gündoğdu F, Baraçlı H. “Evaluation of Potential Locations for Hydropower Plants by Using a Fuzzy Based Methodology Consists of Two-Dimensional Uncertain Linguistic Variables”. Journal of Information Science & Engineering, 38(5), 1-13, 2022.
42. [42] Zadeh LA. “Fuzzy sets”. Information and Control, 3(8), 338-353, 1965.
43. [43] Torra V. “Hesitant fuzzy sets”. International Journal of Intelligent Systems, 529-539, 2010.
44. [44] Atanassov KT. “Intuitionistic Fuzzy Sets”. VII ITKR Session, Sofya, 1983.
45. [45] Zadeh LA. “A note on Z-numbers”. Information Sciences, 181(14), 2923-2932, 2011.