Estimating “good” variability in speech production using invertible neural networks (2020)

• This project was to explore the possibility of using flow-based invertible neural networks for modeling “good” variability space (or the uncontrolled manifold) in speech produciton (Link)

Uncontrolled Manifold Simulator 01 (June 2019)

• This simulates three-joint robot arms and how the UCM can be visualized accordingly (Link).

Uncontrolled Manifold Simulator 02 (Jan 2019)

• Uncontrolled manifolds (UCM) are crucial for human motor learning. The concept is intuitive, but not the implementation. This demonstrates the movement of three-joint robot arms interactively and its null space as a spiral shape (Link).