Ke Sun

Machine learning and beyond

GPG public key
OrcID
Codes on github
Google Scholar
Research Gate

News

Dec 2017I will join the program committee of ICML/IJCAI 2018.
Dec 2017My coordinate system is now centered at Australia.
Oct 2017 I won a prize.

Bio

B.S. in CS (Tsinghua, 2005); M.S. in Math (Tsinghua, 2007); PhD in CS (UniGE, 2015). After some traveling, I am now affiliated with Data61.


Research

I study machine learning (sub-discipline of artificial intelligence). My focus is on the geometric approaches, including a geometry of the observed data, and a more intrinsic geometry of information. I try to build connections between these abstract theories and machine learning. I have recently shifted to unsupervised deep learning.


Recent Works

(see my GScholar page for a manually maintained full list of publications)
  • [link] F. Nielsen and K. Sun. q-Neurons: Neuron Activations based on Stochastic Jackson's Derivative Operators. 2018.

  • [homepage][link] K. Sun and F. Nielsen. Relative Fisher Information and Natural Gradient for Learning Large Modular Models. ICML 2017. (Special thanks to IMLS for partially funding the trip.)

  • [link]F. Nielsen and K. Sun. Clustering in Hilbert simplex geometry. Technical report. 2017.

  • [link]K. Sun and X. Zhang. Coarse Grained Exponential Variational Autoencoders. Technical report. 2017.

  • [link]F. Nielsen, K. Sun and S. Marchand-Maillet. On Hölder Projective Divergences. Technical report. 2017.

  • [homepage][link] F. Nielsen and K. Sun. Guaranteed Bounds on Information-Theoretic Measures of Univariate Mixtures Using Piecewise Log-Sum-Exp Inequalities. Entropy 2016, 18(12), 442.