Ke SUN, PhD
[email]
Thinker

Send physical stuff to

Ke Sun
5/13 Garden St
Eveleigh NSW 2015
Australia

Office: L5-27 (5th floor ATP)
Voice +61 29 490 5972
Voice and/or face "sunk.cs" (Skype)

[me]

GPG public key
OrcID 0000-0001-6263-7355
Research Gate
Google Scholar
LinkedIn

Feb 2019I am an outstanding reviewer of ICLR 2019.
2019Reviewing: ICML 2019, IJCAI 2019, UAI 2019, NeurIPS 2019

Bio

After childhood in Inner Mongolia, I went to Tsinghua University in Beijing, where I got a B.S. degree (2005) from department of computer science and technology and a M.S. degree (2007) from department of mathematical sciences. In the period 2006 to 2009, I discontinuously worked in IBM Research (China) as a part-time contractor (0.5 years) and Chinese University of Hong Kong as a research assistant (1.7 years). From late 2010 to 2015, I did a PhD in computer science in Département d'informatique, Université de Genève in Switzerland. From 2016 to 2017, I was a postdoc in École Polytechnique in Paris and then in King Abdullah University of Science and Technology in Saudi Arabia. Since late 2017 I am a principal research scientist in Data61 (former NICTA), Australia, working closely with the MLRG. In 2017 I received the "Prix de l'excellence de l'UNIGE" from University of Geneva. I have been reviewing for mainstream machine learning conferences (ICML, NeurIPS, etc.).

Research

I study machine learning, a sub-discipline of artificial intelligence. My general interest lies in the spectrum of unsupervised learning, Bayesian statistical learning, manifold/metric learning, and deep learning. I formulate statistical learning problems from the perspective of information geometry (think of general relativity as the curvature of spacetime; information geometry is about the curvature of information), which tackles the essence of information using elegant mathematical tools. A goal of my research is to build a geometric theory of intelligence. I like simple and profound methodologies that are (anti-)intuitive, and therefore my works are more conceptual than practical. Recently I have been focusing on machine learning on graph-structured data. I do not give presentations with explicit rules to avoid equations.

Some Recent Works

(check my Research Gate for a more complete list of publications)
  • [link] F. Nielsen and K. Sun. On the Chain Rule Optimal Transport Distance. arXiv:1812.08113 [cs.LG]. 2018.

  • [web][link] F. Nielsen and K. Sun. Clustering in Hilbert simplex geometry. arXiv:1704.00454 [cs.LG]. 2017. (revised)

  • F. Nielsen and K. Sun. Clustering in Hilbert’s Projective Geometry: The Case Studies of the Probability Simplex and the Elliptope of Correlation Matrices. Geometric Structures of Information. Springer International Publishing. 2018.

  • [link] K. Sun. Intrinsic Universal Measurments of Nonlinear Embeddings. arXiv:1811.01464 [cs.LG]. 2018. (work in progress)

  • U. Akujuobi, K. Sun and X. Zhang. Mining top-$k$ Popular Datasets via a Deep Generative Model. IEEE BigData, 2018. (to appear)

  • M. Avalos, R. Nock, C. S. Ong, J. Rouar and K. Sun. Representation Learning of Compositional Data. NeurIPS 2018. (to appear)

  • [link][arXiv] F. Nielsen and K. Sun. Guaranteed Deterministic Bounds on the Total Variation Distance between Univariate Mixtures. MLSP, Aalborg, Denmark, 2018.

  • [link] F. Nielsen and K. Sun. $q$-Neurons: Neuron Activations based on Stochastic Jackson's Derivative Operators. 2018. arXiv:1806.00149 [cs.NE].

  • K. Sun, F. Malliaros, F. Nielsen, M. Vazirgiannis. Reconstructing Uncertain Graphs Based on Low-Rank Factorizations. Entropy 2018. (poster)