Visible to the public Postdoc and PhD positions in Assured Machine Learning, NTU, SingaporeConflict Detection Enabled

No replies
arvinde's picture
Established Community Member
Joined: Jan 23 2018

Project Description: Over the last few years, Artificial Intelligence (AI) systems have achieved super-human performance in specific yet complex tasks across diverse environments (e.g., image recognition, language translation, complex games like Go). Given their effectiveness, we aim to employ AI (or Agent) Training Programs (ATPs) to generate scenarios automatically for training (on specific tasks) in safety-critical applications while addressing the issue of trust to improve adoption:

Trust to Train: Gain trust of organizations employing ATPs through guarantees on training outcomes (safety), equality in training for all trainees (fairness), and training for unexpected situations (robustness).

Train to Trust: Explainable (understandable) and effective feedback interface in training to gain trust of trainees.

We intend to develop and assess Explainable and tRustworthy (ExpeRt) ATPs with feedback interfaces to adaptively train human(s) for safety-critical applications with show-case projects on emergency response and maritime navigation.

Team: This project is a collaboration between faculty at Singapore Management University (SMU), National University of Singapore (NUS) and Nanyang Technological University (NTU).

* Pradeep Varakantham, research interests in intelligent agent systems that perform dynamic matching of supply and demand

* Akshat Kumar, research interests in the intersection of AI and ML with a focus on multiagent decision making and reinforcement learning

* Arunesh Sinha, research interests in game theory, machine learning, and security.

* Arvind Easwaran, research interests in the design and analysis of safety and time-critical cyber-physical systems

* David Lo, research interests in the intersection of software engineering, cybersecurity and data science

* Harold Soh, research interests in human-AI/robot interaction, machine learning and robotics

* Vivian Hsueh-Hua Chen, research interests in gamification for social well being

Job Opportunities: Within the trustworthiness assurance research track in the above project, we will be exploring black-box and grey-box techniques for assessing the safety, robustness and fairness properties in deep reinforcement learning frameworks used in the ATPs. In particular, we are interested in scalable techniques with statistical guarantees, which can handle multi-stage interactions with the ATPs. For this purpose, we are looking for PhD students and post-doctoral candidates with interest and/or experience in the following areas:

* Black-box and grey-box testing methods for deep neural networks, particularly deep reinforcement learning methods.

* Probably approximately correct (PAC) learning framework for deriving probabilistic guarantees on the above black-box and grey-box methods.

It is expected that potential candidates have a strong interest and/or experience in theoretical aspects related to the PAC framework.

If you are interested in applying for any of these positions, please send your CV to