Mahdi Kazemi Moghaddam
Artificial Intelligence and Machine Learning Researcher

About Me


I am a postdoctoral research fellow at Deakin University working with Prof. Richard Dazeley and Prof. Peter Vamplew. My research is about Multi-Agent Systems (MAS), human robot interaction, and Reinforcement Learning (RL).

Previously, I was a visiting researcher at the Autonomous Agents Research Group, University of Edinburgh, where I was working with Assistant Prof. Stefano V. Albrecht, on mutli-agent RL for autonomous driving.

I obtained my PhD from the University of Adelaide where I was working with Prof. Javen Shi and Associate Prof. Qi Wu, on single-agent and multi-agent RL for real-world problems such as visual navigation. During my PhD, I was associated with the Australian Institute for Machine Learning (AIML) and the Australian Center for Robotic Vision.

Besides research, I, also, have extensive experience in building computer vision and machine learning tools to address different industry problems such as sports analysis from videos, bushfire mitigation from satellite imagery, shopping behaviour analysis from surveillance videos, etc.

In my free time, I enjoy swimming and driving, besides reading a paper from the ever-growing ordered list of papers to read!

Interests


My personal interests include:

  • RL. Learning from trial and error/ interaction with the environment is a facsinating area of research. This is how humans learn to survive from day 1!
  • Embodied AI. Having a physical embodiement for the AI agent makes me feel like we are getting closer to creating human-like intelligence.
  • Computer Vision. Intelligence heavily depends on perception. Our embodied agents need to first percieve the world around them, mainly from visual sensory inputs, in order to then make the right decisions.

Industry Relations

  • Different industry projects I work on multiple industry projects to help fast track the positive impact AI/ ML can have on our daily lives. I am currently engaged in industry projects to help mitigate the bushfire impacts using AI/ ML, help table tennis coaches with ML-based video analysis among others.

News


Highlights

  • Nov 2022: I have successfully completed my PhD from the University of Adelaide. My thesis titled "Towards Optimistic, Imaginative, and Harmonious Reinforcement Learning in Single-Agent and Multi-Agent Environments" is publicly accessible here .
  • Oct 2022: I have joined Deakin University as a Postdoctoral Research Fellow in Reinforcement Learning.
  • Apr 2022: I have joined Autonomous Agents Research Group at the University of Edinburgh, as a visiting researcher.
  • Nov 2021: my paper titled "ForeSI: Success-Aware Visual Navigation Agent" is accepeted for publication at WACV2022. Link to paper
  • Jul 2021: my paper/ poster titled "Success-Aware Visual Navigation Agent" is accepeted at CVPR2021 Embodied AI workshop. Link to poster
  • Jun 2021: our NOBURN Citizen Science funding (~0.5 M) is approved. Privilaged to have co-authored the grant proposal and be a machine learning advisor of the project.
  • Nov 2020: my paper titled "Optimistic Agent: Accurate Graph-Based Value Estimation for More Successful Visual Navigation" is accepeted for publication at WACV2021. Link to paper
  • Nov 2020: I won 1st place in the ACRV's Carlie Showcase robotic challenge and 2nd place in Carlie Vision robotic challenge.
  • Sep 2020: I, as part of a team of five including two bushfire experts and two other data scientists, was one of the four finalist teams in the DataQuest2020 Bushfire challenge.
  • Aug 2019: I was awarded the Adelaide Graduate Research Scholarship to start my PhD in machine learning, at the Australian Institute for Machine Learning (AIML).
  • Jul 2019: I was awarded the Valedictorian of class 2019 (single top-performing student among 8 schools in the faculty of ECMS), and the Dean's Academic Excellence Award.
  • Dec 2018: I was awarded the Summer Research Scholarship to do research in computer vision for shopping video analysis.