Ph.D. Thesis: Adaptive Multi-Objective Reinforcement Learning for Traffic Signal Control Based on Cooperative Multi-Agent Framework (2010 - 2013)
- Develop a reinforcement learning based traffic signal control framework.
- Combine the strong sequential decision making of the dynamic programming with the adaptation power of reinforcement learning examining the hybridization of dynamic programming and machine learning.
- Develop a notion of optimality by converting a multi-objective control function into a single-objective function.
- Use multi-agent system, in a cooperation-based configuration, to comply with the distributed nature of the problem and overcome its inherent complexity.
M.Sc. Thesis: Designing Multi-Agent Unit Tests Using Systematic Test Design Patterns (2005 - 2010)
- Design test cases based upon unit testing approach for MASs through an Eclipse Plug-in.
- Aim at removing faults along the application development.
- Help MASs developers in testing each agent individually.
- It relies on the use of Mock Agents to guide the design and implementation of agent unit test cases.
- The implementation of this approach is built on top of Java Agent Development Environment (JADE) platform.
- Extend the JUnit test framework to allow the execution of JADE unit test cases.
- Provide test stereotypes implementations for a vast majority of agent design patterns.
Graduation Project: An Agent-Based Bidding System: Implementing an Electric Power Market (2004 - 2005)
- Implement an agent-based auction for buying and selling electric power;
- the original project link: http://www.aesc-inc.com/download/epri.pdf
- Information is provided to all participants via the auction website (server);
- each participant is represented by an agent who is either an auctioning or a bidding agent (client).
- Implementation tools and technologies:
- Database: XML files, Agent tools: Java Agent Development Framework (JADE), Web Server: Apache Tomcat, IDE: Jbuilder, Web Forms: JSPs and HTML.