Hi! I'm Gary Ang, a rather atypical (or rather relatively old) PHD in Computer Science candidate at the Singapore Management University. Prior to this mid-life break, I headed the investment risk management function at the Monetary Authority of Singapore. I am also an avid illustrator who has been featured in various showcases and worked with Wallpaper and Jetstar Asia (more at PLAYGRD)


I am interested in research relating to deep learning on networks (or graphs), specifically dynamic and/or temporal networks with multimodal attributes. My interests in multimodal dynamic networks stems from my parallel interests in the creative and financial domains, where such networks are common.


In the creative domain, different design or creative objects (e.g., user interface elements, graphical design elements) may be connected to each other via different types of relationships at different levels. Psychological evidence shows that humans parse images into part-whole hierarchies and model the viewpoint invariant spatial relationships between a part and a whole as they process a piece of visual information (Hinton, 2021). Such relationships are a core part of crative works since the part-whole hierarchical relationships influence how one understands, interacts and experiences creative works.


One of my works in this area, on Learning Network-Based Multi-Modal Mobile User Interface Embeddings which has been published at the 26th International Conference on Intelligent User Interfaces (IUI 2021) proposes a novel self-supervised model - Multi-modal Attention-based Attributed Network Embedding (MAAN) model - that learns semantically rich information from networks of UI design objects with multiple modalities - text, code, images, categorical and numerical data. [Paper]


In the financial domain, different economic entities (e.g., countries, companies, individuals) may be connected to each other via different types of financial relationships (e.g., transactions). Networks in the financial domain vary across time - economic entities appearing and disappearing; shifting relationships between economic entities; and time-varying attributes associated with the economic entities and/or their relationships from multiple modalities (market prices, news and social media). Hence, financial networks are particularly suited for research on network models that learn multimodality and dynamicity in networks.


One of my works in this area, on Learning Knowledge-Enriched Company Embeddings for Investment Management which has been accepted by the 2nd ACM International Conference on AI in Finance (ICAIF 2021) proposes Knowledge-Enriched Company Embedding (KECE), a novel multi-stage attention-based dynamic network embedding model that combines multimodal information of companies with knowledge from Wikipedia and knowledge graph relationships from Wikidata to generate company entity embeddings that can be applied to a variety of downstream investment management tasks. Paper will be made available soon.