Report on the Neural Networks in Econometric System Identification, Forecasting and Control workshop
The 3-day MEAFA research training workshop on Neural Networks in Econometric System Identification, Forecasting and Control, 15-17 Feb 2010, confirmed all the expectations and counted with the presence of 22 people from the academia (PhD students and academics from a wide range of disciplines) and private sector.
The workshop was presented by Dr Hans Georg Zimmermann who is the Senior Principal Research Scientist at Siemens AG - Corporate Technology in Germany (and responsible for an international team of more than 2,000 researchers).
The workshop was fully subsidised for MEAFA members and graduate and higher research degree students from the University of Sydney, whereas academic members were asked to cover the daily operating cost for their participation. Participants included also practitioners from the private sector.
Over the 3-day workshop a wide range of concepts and applications of neural networks was covered. From almost 20 years researching in this area, Dr Zimmermann made evident all his experience during the workshop.
During the first session, Dr Zimmermann reviewed the basic concepts and literature and then proceeded to to more complex systems of identification, forecasting and control models. Dr Zimmermann defended the role of neural networks as an important method that uses standard econometric tasks to achieve optimal decision making. The question of whether neural networks provides new methodological insights and better understanding of the optimal decision making process, as well as the limitations of the methodology were expanded during the course. Dr Zimmermann offered a more academic dimension in conjunction with practical examples, where he had personally applied neural networks to solve real world problems while working at Siemens.
The workshop focused in understanding the appropriate neural network architectures and corresponding learning algorithms, which are beyond pure data-driven modelling, and where many types of applications such as long term forecasting, or the control of dynamical systems can be tackled. Even in the finance world, where markets are highly interrelated and the analysis of an open dynamical system is not sufficient, Dr Zimmermann demonstrated that the neural network framework should be considered a new approach to risk analysis.
We thank all attendants for their support and active participation in the discussions.