In recent years there has been much attention towards identification of dynamically interconnected systems like biological networks, cyber-physical systems and economic networks. One main question in network identifications is to identify a system embedded in the dynamic network with limited use of measured and excitation signals and with maximum accuracy. I am developing techniques to identify a system embedded in a dynamic network with maximum accuracy, also developing conditions under which it can be met which includes sensor allocation and actuator positioning in dynamic networks.
Kernel methods and regularization are well developed tools in the field of Machine learning. Recently, regularized kernel-based methods have been introduced to identify simple linear dynamic systems. Based on these methodologies, I am developing techniques to identify large scale interconnected dynamic systems which provides a significant advantage in the field of dynamic networks.
During my Masters I worked on Oil and gas reservoir systems. My focus was on methods to find optimal control strategies to effectively and efficiently extract oil/gas from the reservoir. My master thesis was on finding effective sampling strategies to calculate the stochastic gradient which is used to find the optimal input strategy to extract oil in reservoirs.