• Risk-Reduction in Markov Decision Processes (MDPs)
  • Factors leading to Entrepreneurial Risk in a SoS and Project Success
  • Using Simulation to Improve the Layouts of a Healthcare Facility
  • Reinforcement Learning for Remanufacturing

Risk-Reduction in Markov Decision Processes (MDPs)


Risk-Reduction in Markov Decision Processes (MDPs)


PROJECT DESCRIPTION
The goal of this project is to study a new class of algorithms in the area of MDPs that seek to adjust for risk and at the same time maximize potential revenues. We have had encouraging results with two new algorithms developed: one for reducing variance and the other for downside risk.


PUBLICATIONS

  1. Solving Markov Decision Processes with Downside Risk Adjustment,” Gosavi, A. and A. Parulekar.  To appear in the International Journal of Automation and Computing
  2. Variance-penalized Markov decision processes:Dynamic programming and reinforcement learning techniques,” Gosavi, A. International Journal of General Systems, 43(6), 649-669, 2014.
  3. Beyond Exponential Utility Functions: A Variance-Adjusted Approach for Risk-Averse Reinforcement Learning,” Gosavi, A., S.K. Das, and S.L. Murray. Proceedings of IEEE, SSCI, ADPRL, Orlando, FL, 2014. 

Factors leading to Entrepreneurial Risk in a SoS and Project Success


Factors leading to Entrepreneurial Risk in a System of Systems (SoS) and Project Success


PROJECT DESCRIPTION
The goal of this project was (i) to study mechanisms to predict the behavior of an individual system within an SoS from the perspective of the central controller and (ii) identify factors that can help improve the success of a defense project with numerous players who are not co-located. This resulted in the following papers (this was partly funded by a grant from DoD).


PUBLICATIONS

  1. Predicting Response of Risk-Seeking Systems during Project Negotiations in a System of Systems,” Gosavi, A., S. Agarwal, and C. H. Dagli. To appear in the IEEE Systems Journal.
  2. Attitudes towards Face-To-Face Meetings in Virtual Engineering Teams: Perceptions from a Survey of Defense Projects,” Blenke, L. A. Gosavi, and W. Daughton. To appear in the International Journal of Project Organisation and Management.
  3. Flexible and Intelligent Learning Architectures for SoS (FILA-SoS): Architectural Evolution in Systems-of-Systems,’ Agarwal, S., L. E. Pape, C. H. Dagli, N. K. Ergin, D. Enke, A. Gosavi, R. Qin, D. Konur, R. Wang, R. D. Gottapu. Procedia Computer Science, Conference on Systems Engineering Research (CSER), Hoboken, NJ, 2015.

Using Simulation to Improve the Layouts of a Healthcare Facility


Using Simulation to Improve the Layouts of a Healthcare Facility


PROJECT DESCRIPTION
Healthcare facilities are undergoing a dramatic change in order to accommodate the patient-centered theme that most facilities are now embracing. This resulted in the following papers (and this was partly funded by a grant from the VA Hospitals).


PUBLICATIONS

  1. Analysis of Clinic Layouts and Patient-Centered Procedural Innovations Using Discrete-Event Simulation,” Gosavi, A., E. Cudney, S.L. Murray, and S. Masek.  To Appear in the Engineering Management Journal.
  2. Improving flow in health care clinics through simulation,” Cudney, E., D. Scroggins, S. Murray, and A. Gosavi.   Proceedings of IIE Annual Conference, Orlando, 2012.
  3. “The use of discrete-event simulation to improve flow in healthcare clinics,” Cudney, E., D. Scroggins, S. Murray, and A. Gosavi. Proceedings of the Annual ASEM Conference, Virginia Beach, VA, 2012.

Reinforcement Learning for Remanufacturing


Reinforcement Learning for Remanufacturing


PROJECT DESCRIPTION
This is an ongoing project in which we seek to use Reinforcement Learning Techniques to solve challenging problems from the domain of environmentally conscious green manufacturing. Many problems in remanufacturing industries can be set up as Markov decision processes where RL can be a useful solution method.


PUBLICATIONS

  1. How to Rein in the Volatile Actor: A New Bounded Perspective,” Gosavi, A.  Procedia Computer Science, Complex Adaptive Systems Conference, Philadelphia, PA, Volume 4, 2014.
  2. Relative Value Iteration for Average Reward Semi-Markov Control via Simulation,” Gosavi, A.  Proceedings of the Winter Simulation Conference, Washington DC, December 2013.