UFAZ Alumnus Presented Paper at International Conference

Javidan Abdullayev, a graduate of the Master’s program in Data Science and Artificial Intelligence at the French-Azerbaijani University (UFAZ), participated in the IJCNN 2025 conference.

Held from June 30 to July 5, the event brought together experts in neural networks and artificial intelligence in Rome, at the Pontifical Gregorian University. The conference, themed “All Neural Network Roads Lead to Rome”, highlighted the central role of neural networks across various scientific fields and practical applications.

As part of the program, special sessions and workshops were organized on supervised and unsupervised learning, reinforcement learning, convolutional and spiking neural networks, cognitive algorithms, deep learning, and creative applications. Javidan Abdullayev presented his paper titled “Enhancing Time Series Classification with Diversity-Driven Neural Network Ensembles”.

The paper introduced a novel approach to promote feature diversity within neural network ensembles. To achieve this, a feature orthogonality loss was applied, ensuring that each model learns complementary representations. As a result, the ensemble produces richer features, achieves stronger performance, and attains state-of-the-art accuracy across 128 time series datasets. This approach is also more efficient compared to traditional ensemble methods.

We congratulate our graduate on this achievement and wish continued success and impactful contributions in his future research!