The article titled “Multiplicative Gradient Decomposition: A Hybrid Additive–Multiplicative Optimization Framework,” authored by Asst. Prof. Sajedeh Norozpour SIGAROODI from the Civil Engineering Department, has been published in the Q1-indexed journal Mathematics.
The study proposes a hybrid additive–multiplicative optimization framework for training machine learning models. Unlike conventional optimization methods, the proposed approach combines additive and multiplicative parameter updates, enabling the simultaneous modeling of both linear and exponential dynamics.
The developed “Hybrid Multiplicative Gradient Decomposition (HMGD)” framework demonstrated faster convergence, improved stability, and stronger robustness against multiplicative noise, particularly in nonlinear optimization problems and systems involving multiplicative interactions. The study is also expected to contribute to the development of new optimization approaches in machine learning and computational mathematics.
The department congratulates Asst. Prof. Sajedeh Norozpour SIGAROODI on this valuable scientific contribution and wishes her continued success in her academic work.
