Continuous Riemannian Geometric Learning for Sleep Staging Classification

Seungwoo Jeong, Wonjun Ko, Ahmad Wisnu Mulyadi, Heung-Il Suk

Abstract: Sleep staging classification has recently received a lot of attention because of its importance and has shown remarkable achievements through deep neural models, but lacks consideration of geometrical structure or continuous time. In this paper, we propose to exploit a diffeomorphism mapping between Riemannian manifolds and a Cholesky space. Further, in order for continuous modeling, we devise a continuous manifold learning method by integrating a manifold ordinary differential equation and a gated recurrent neural network. We demonstrate the validity of our proposed method through experiments using a publicly available SleepEDF-20 dataset.