Vivienne Huiling Wang

My research sits at the intersection of reinforcement learning, probabilistic and generative modelling, and interactive multi-agent learning, aiming to build agentic AI systems that can plan, act, remember, and cooperate reliably under uncertainty. Selected papers and patent are listed as follows. Full list of publications can be found in my Google Scholar page.

A GP-regularized hierarchical diffusion policy models uncertainty to generate subgoals, improving sample efficiency and control performance.

We propose Multi-Scale Fusion (MSF) to enhance VAE guidance for Object-Centric Learning (OCL) training.

We propose simple yet effective unified architecture for OCL, which outperforms baselines in object discovery and recognition, as well as downstream visual prediction and reasoning.

We propose a new Gaussian processes (GPs) based method for learning probabilistic subgoal representations in Hierarchical Reinforcement Learning (HRL).

We propose a novel adversarially guided subgoal generation framework for goal-conditioned HRL to mitigate the issue of non-stationarity in off-policy training.

We address semantic video object segmentation via a novel cross-granularity hierarchical graphical model to integrate tracklet and object proposal reasoning with superpixel labeling.

We present a novel algorithm that detects recurring primary object and learns cohort object proposals over space-time in video. Our core contribution is a graph transduction process that exploits both appearance cues learned from rudimentary detections of object-like regions, and the intrinsic structures within video data.

We propose a semi-supervised approach to adapting CNN image recognition model trained from labeled image data to the target domain exploiting both semantic evidence learned from CNN, and the intrinsic structures of video data.

We combine geodesic distance information with the flexibility of level set methods in energy minimization, leveraging the complementary strengths of each to promote accurate boundary placement and strong region connectivity while requiring less user interaction.