CV | Research Statement | Github
Google Scholar | DBLP | ResearchGate




Current Affilation

Empirical Inference Department
Max Planck Institute for Intelligent Systems
Max-Planck-Ring 4, 72076 Tübingen, Germany
Tel. +49 (0)7071 601 559
Fax. +49 (0)7071 601 552

My Books

Kernel Mean Embedding of Distributions: A Review and Beyond
K. Muandet, K. Fukumizu, B. Sriperumbudur and B. Schölkopf
Foundations and Trends in Machine Learning: Vol. 10: No. 1-2, pp 1-141.


Hallo! My name is Krikamol Muandet (ไกรกมล หมื่นเดช). I am a research group leader affiliated with the Empirical Inference Department at Max Planck Institute for Intelligent Systems, Tübingen, Germany. From January 2016 to December 2017, I was a lecturer at the Department of Mathematics, Faculty of Science, Mahidol University in Thailand. My research interest lies in the area known as "machine learning". I am particularly interested in, for example, statistical learning theory, kernel methods, Bayesian nonparametric, large-scale learning, and counterfactual prediction (see a full list of my publications for details). When I am not doing research, I enjoy reading books (on topics related to philosophy, psychology, history of science, etc) and watching movies. I also like outdoor sports such as bouldering, climbing, and snowboarding (if the weather permits).

My current research aims to develop machine learning techniques that will bridge the gap between randomized experiments and empirical inference, enabling machines to better infer causality from data. It has applications in observational studies, medical diagnosis, economics, and online advertisement, for example.

As a PhD student, I have worked primarily with Prof. Bernhard Schölkopf. In December 2015, I was awarded the doctoral degree with summa cum laude, that is, "with greatest honor", from the University of Tübingen. I previously obtained a master's degree with distinction in machine learning from University College London (UCL), United Kingdom. At UCL, I worked primarily with Prof. Yee Whye Teh. (M.Sc. thesis advisor) at the Gatsby Computational Neuroscience Unit and Prof. John Shawe-Taylor (M.Sc. Tutor) at the Center for Computational Statistics and Machine Learning. During my PhD, I was a visiting scholar at the Institute of Statistical Mathematics, Japan; Center for Cosmology and Particle Physics, New York University; Palomar Observatory in San Diego; American Museum of Natural History, and Institut für Stochastik und Anwendungen, University of Stuttgart, among others.

In 2011, it was a great honour for me to co-organize a Festschrift symposium to honor Prof. Vladimir Vapnik, on the occasion of his 75th birthday. I also helped organize the 29th Neural Information Processing Systems (NIPS 2016) and Data, Learning, and Inference (DALI 2019) workshop. In December 2016, I was also invited to participate in the Dagstuhl Seminar in New Directions for Learning with Kernels and Gaussian Processes. At the seminar, we discussed various prospects of kernel methods in machine learning.

I always seek new collaboration. If you are interested in working with me, I will be glad to hear from you.


Here are some higlights of my recent work. For the complete information, see my full list of publications.

Local Temporal Bilinear Pooling for Fine-grained Action Parsing
Zhang, Y., Tang, S., Muandet, K., Jarvers, C., Neumann, H.
IEEE Computer Society Conference on Computer Vision and Pattern Recognition (CVPR2019)
ArXiv preprint:

Summary: In this paper we propose a novel bilinear pooling operation, which is used in intermediate layers of a temporal convolutional encoder-decoder net. In contrast to other work, our proposed bilinear pooling is learnable and hence can capture more complex local statistics than the conventional counterpart. In addition, we introduce exact lower-dimension representations of our bilinear forms, so that the dimensionality is reduced with neither information loss nor extra computation. We perform intensive experiments to quantitatively analyze our model and show the superior performances to other state-of-the-art work on various datasets.

Counterfactual Mean Embedding: A Kernel Method for Nonparametric Causal Inference
K. Muandet, M. Kanagawa, S. Saengkyongam and S. Marukatat
ArXiv preprint:

Summary: In this paper, we proposes a novel Hilbert space representation of a counterfactual distribution---called counterfactual mean embedding (CME)---with applications in nonparametric causal inference. To infer the outcomes of certain interventions, we propose to embed the counterfactual distribution into a reproducing kernel Hilbert space (RKHS). Under appropriate assumptions, the CME allows us to perform causal inference over the entire landscape of the counterfactual distribution. We apply the proposed estimator to off-policy evaluation tasks to demonstrate its advantages.



I feel really fortunate to have worked with these wonderful people.

  • Bernhard Schölkopf (MPI-IS, Germany)
  • Kenji Fukumizu (ISM, Japan)
  • Bharath Sriperumbudur (PennState, USA)
  • Yee Whye Teh (Oxford & DeepMind, UK)
  • Kun Zhang (CMU, USA)
  • David Balduzzi (DeepMind, UK)
  • Arthur Gretton (Gatsby Unit, UCL, UK)
  • Francesco Dinuzzo (Amazon)
  • Ilya Tolstikhin (MPI-IS, Germany)
  • David Lopez-Paz (Facebook AI Research)
  • Zhikun Wang (Google)
  • Stefan Harmeling (Dusseldorf)
  • Jonas Peters (U of Copenhagen, Denmark)
  • Sanparith Marukatat (NECTEC, Thailand)
  • Cholwich Nattee (SIIT, Thailand)
  • David Hogg (NYU, USA)
  • Rob Fergus (Facebook and NYU, USA)
  • Motonobu Kanagawa (MPI-IS, Germany)
  • Sorawit Saengkyongam (Agoda, Thailand)
  • To be updated soon!