ɫɫ��

Biosketch 

Fong holds four degrees in CS and EE from three universities: Imperial College London, University of Oxford, and University of Auckland. His research activities revolve around next-generation artificial intelligence (NG-AI). Specific focus areas include data-driven and goal-informed knowledge discovery, ontological knowledge representation and reasoning, and applied machine learning. His publications and intellectual property contributions include two books, 13 book sections, 100 journal papers, 100 conference papers, and two international patents. Archival journals that carry his work include IEEE T-KDE, IEEE T-AC, IEEE T-II, IEEE T-EC, and several other IEEE Transactions titles. He has been an Associate Editor of IEEE T-CE since 2013. He has recently served as a guest editor of IEEE CE Magazine’s special section on Machine Learning for End Consumers, which will appear in 2020 (DOI 10.1109/MCE.2020.2986934). Dr. Fong is a Fellow of IET, European Engineer (Eur Ing), and Chartered Engineer (CEng).

Research

Applied Artificial Intelligence

Data-driven, goal-informed AI can be used for optimal resource allocation and discovery of hidden knowledge in data while keeping the size of the hypothesis space relatively manageable. Focus areas include applied machine learning, ontological knowledge representation and reasoning, and perception. Depending on the data sources and use cases, the general approach is applicable to a wide range of real-world problems across multiple domains, e.g. healthcare, business and commerce, industry, education, and scientific discovery

Related Publications (recent / selected)

• Eslami T, Mirjalili V, Fong A, Laird A, and Saeed F, ASD-DiagNet: A hybrid learning approach for detection of autism spectrum disorder using fMRI data, Frontiers in Neuroinformatics, Vol. 13/70, 11 pages, 2019. DOI: 10.3389/fninf.2019.00070
• Qolomany B, Al-Fuqaha A, Gupta A, Benhaddou D, Alwajidi S, Qadir J, and Fong A, Leveraging machine learning and big data for smart buildings: a comprehensive survey, IEEE Access, Vol. 7/1, pp. 90316-90356, December 2019. DOI: 10.1109/ACCESS.2019.2926642
• Afzaal M, Usman M, and Fong A, Tourism mobile app with aspect-based sentiment classification framework for tourist reviews, IEEE Trans. Consumer Electronics, Vol. 65/2, pp. 233 - 242, 2019. DOI: 10.1109/TCE.2019.2908944
• Nguyen ML, Hui SC, and Fong A, Submodular memetic approximation for multiobjective parallel test paper generation, IEEE Trans. Cybernetes, Vol. 47/6, pp. 1562–1575, 2017. DOI: 10.1109/TCYB.2016.2552079
• Azeem M, Usman M, and Fong A, A churn prediction model for prepaid customers in telecom using fuzzy classifiers, Telecommunication Systems, Vol. 66/4, pp. 603–614, 2017.
• Tang P, Hui SC, and Fong A, A lattice-based approach for chemical structural retrieval, Engineering Applications of Artificial Intelligence, Vol. 39, pp. 215–222, 2015. http://dx.doi.org/10.1016/j.engappai.2014.12.006

Toward Next-Generation Artificial Intelligence

Next-generation artificial intelligence (NG-AI) centers around the idea of contextual awareness and human-centric computing. Focus areas include commonsense reasoning and question-and-answer agents. Both areas are underpinned by advances in natural language processing. The work toward NG-AI is spearheaded by two current graduate research students.

Sirwe Saeedi (graduate student): “Machine Learning and Natural Language Processing for Commonsense Reasoning”

Currently, we are working on a project aimed at developing machine learning algorithms to understand human commonsense reasoning using natural language processing techniques. We are investigating the use of deep neural networks and state-of-the-art architectures to implement a pipeline for commonsense reasoning tasks using, for example, BERT, ULMFiT, and RoBERTa with Hugging Face transformers. We have made significant progress in our research work that will enable us to participate in SemEval-2020, a major global competition in natural language processing and semantic evaluation. We are in the process of summarizing our findings in an article that will be submitted to the SemEval-2020 conference.

Rabeya Bibi (graduate student): “Open Domain Question Answering: A Benchmark Task in Natural Language Interpretation”

General-purpose, open-domain intelligent question and answer (QA) agents remain an active area of research despite widespread use of commercial QA products. There is still a great deal to be done to accurately and efficiently fulfill QA task in natural language understanding or natural language interpretation. The approach taken in this line of research is one of divide and conquer: begin by working with modular units or subtasks and then gradually move towards finding an optimal solution. In this regard, identifiable subtasks include: finding a suitable question, intent classification of answer with respect to the selected question, exploring and mitigating the gap between synthesized and human-generated QA pairs, finding an optimal tradeoff to question generation and the span of answer, e.g. sentence-level, paragraph-level, across-documents, etc. Finally, developing appropriate evaluation metrics and methods for the subtasks is as crucially important as finding the upper bounds of these metrics.