MACHINE LEARNING APPROACHES FOR INTERPRETING VISUAL DATA UNDER CONDITIONS OF UNCERTAINTY
DOI:
https://doi.org/10.35546/kntu2078-4481.2024.4.31Keywords:
artificial intelligence, computer vision, neural networks, artificial neural networks, convolutional neural networks, recurrent neural networks, pattern recognitionAbstract
In the field of computational intelligence, the interpretation of visual data underlies many applications, from autonomous navigation systems to medical diagnostics. However, the inherent uncertainty present in visual data due to factors such as noise, occlusions, and variability in object appearance pose significant challenges to reliable interpretation. Traditional computer vision methods often struggle with the ambiguity and inaccuracy of real-world data, leading to a demand for more adaptive and robust methods. Machine learning, with its ability to learn from and adapt to complex data patterns, becomes a key solution to this puzzle. This paper explores advanced machine learning approaches designed to interpret uncertain visual data, which is critical to improving the accuracy and reliability of automated visual understanding. The emergence of machine learning in the field of visual data interpretation initiated a paradigm shift from rule-based processing to data-driven learning. Among the many challenges, the uncertain interpretation of visual data requires a fine-grained approach where the system must not only recognize patterns, but also quantify and manage the uncertainty inherent in the input data. This uncertainty can take many forms, including but not limited to sensor noise, partial occlusions, and ambiguous object boundaries, which together degrade the performance of conventional translation systems. Therefore, this research paper aims to analyze various machine learning models and algorithms that are specially developed or adapted for the interpretation of visual data under conditions of uncertainty. It focuses on understanding how these techniques can effectively process and interpret data that is ambiguous, incomplete, or distorted. The purpose of the research is to choose the optimal approach to the analysis of visual data using artificial intelligence under conditions of uncertainty. The main task of the research is the analysis of classification methods based on well-known neural networks, which are focused on the processing and recognition of image patterns.
References
Abhishek G., Alagan A., Ling G., Ahmed S. K. Deep learning for object detection and scene perception in selfdriving cars: Survey, challenges, and open issues. Array. 2021. T. 1, № 10. С. 1-10. DOI: https://doi.org/10.1016/j.array.2021.100057
Arshi P., Muhammad A. K., Rukhsana Z., Huma A., Muhammad A., Muhammad M. F. Vision Transformers in medical computer vision. Engineering Applications of Artificial Intelligence. 2023. Т. 1, № 122. С. 1-10. DOI: https://doi.org/10.1016/j.engappai.2023.106126
Lee M., Valisetty R., Breuer A., Kirk K. Current and Future Applications of Machine Learning for the US Army. URL: https://apps.dtic.mil/sti/pdfs/AD1050263.pdf (дата звернення: 20.10.2024).
Karaman M., Çatalkaya H., Aybar C. Institutional Cybersecurity from Military Perspective. International journal of information security science. 2016. № 5. – C. 1-11. – URL: https://www.researchgate.net/publication/299533127_Institutional_Cybersecurity_from_Military_Perspective
Chang D. T. Bayesian Neural Networks: Essentials. URL: https://www.researchgate.net/publication/353067263_Bayesian_Neural_Networks_Essentials (дата звернення: 20.10.2024).
Rabaey P., De Boom C., Demeester. T. Neural Bayesian Network Understudy. URL: https://www.researchgate.net/publication/365422792_Neural_Bayesian_Network_Understudy (дата звернення: 20.10.2024).
Bonnet D., Hirtzlin T., Majumdar A., Dalgaty T. Bringing uncertainty quantification to the extreme-edge with memristor-based Bayesian neural networks. Nature Communications. 2023. Т. 1, № 14. С. 3-10. DOI: http://dx.doi.org/10.1038/s41467-023-43317-9
Khan Tareen S., Khan Tareen F. Convolutional Neural Networks for Beginners. SSRN Electronic Journal. 2023. С. 1-10. URL: https://www.researchgate.net/publication/373813288_Convolutional_Neural_Networks_for_Beginners 9. Liu C. Research on image classification leveraging deep convolutional neural networks and visual. Applied and Computational Enginering. 2024. Т. 1, № 32. С. 200-209. DOI: https://doi.org/10.54254/2755-2721/32/20230212
Wang R. (2024). Generalisation of Feed-Forward Neural Networks and Recurrent Neural Networks. Applied and Computational Enginering. 2024. Т. 1, № 40. С. 242-246. DOI: https://doi.org/10.54254/2755-2721/40/20230659
Agrwal P., Kumar S., Shekhar Yadav C. Denoising watermarked images using Bidirectional Recurrent Convolutional Neural Networks. URL: https://www.researchgate.net/publication/371899317_Denoising_watermarked_images_using_Bidirectional_Recurrent_Convolutional_Neural_Networks (дата звернення: 20.10.2024).