Scopus Publications
- Challenges in Designing Music Recommendation Systems
Nikita S. Churilov, Iana S. Petrova
Proceedings of the 2026 Elcon Conference of Young Researchers on Computing and Processing and Information Security Elcon CP 2026, 2026 - Comparative Analysis of Usability Testing Automation Systems
Mikhail Y. Kononykhin, Sergei A. Pavlov, Julia A. Vishnevskaya, Iana S. Petrova
Proceedings of the 2026 Elcon Conference of Young Researchers in Electrical Engineering Automation and Control Systems Elcon EE 2026, 2026 - A Comparison of Feature Extraction Models for Images with Multiple Annotations
I. S. Petrova, G. S. Ivanova
Studies in Systems Decision and Control, 2024 - Comparison of Cross-Entropy Loss and Metric Losses in the Multi-Label Classification
Iana Sergeevna Petrova, Galina Sergeevna Ivanova
Proceedings 2023 5th International Conference on Control Systems Mathematical Modeling Automation and Energy Efficiency Summa 2023, 2023
Loss functions for computer vision neural network training in multi-label classification problems are considered. Comparison of cross-entropy loss, triplet loss with different mining strategies, and lifted structured loss is conducted as a result of multiple experiments. K-NN algorithm is applied to classify image embeddings. Parameter selection for the metric-learning model is described. The advantages and disadvantages of metric losses and cross-entropy loss are listed and analyzed, regarding computational costs, recognition accuracy, and the possibility of application in downstream tasks.
