Share This Author
High-Dimensional Feature Selection by Feature-Wise Kernelized Lasso
- M. Yamada, Wittawat Jitkrittum, L. Sigal, E. Xing, Masashi Sugiyama
- Computer ScienceNeural Computation
- 2 February 2012
It is shown that with particular choices of kernel functions, nonredundant features with strong statistical dependence on output values can be found in terms of kernel-based independence measures such as the Hilbert-Schmidt independence criterion and the globally optimal solution can be efficiently computed.
Interpretable Distribution Features with Maximum Testing Power
In real-world benchmarks on high-dimensional text and image data, linear-time tests using the proposed semimetrics achieve comparable performance to the state-of-the-art quadratic-time maximum mean discrepancy test, while returning human-interpretable features that explain the test results.
K2-ABC: Approximate Bayesian Computation with Kernel Embeddings
- Mijung Park, Wittawat Jitkrittum, D. Sejdinovic
- Computer ScienceInternational Conference on Artificial…
- 9 February 2015
This paper proposes a fully nonparametric ABC paradigm which circumvents the need for manually selecting summary statistics, and uses maximum mean discrepancy (MMD) as a dissimilarity measure between the distributions over observed and simulated data.
A Linear-Time Kernel Goodness-of-Fit Test
A novel adaptive test of goodness-of-fit, with computational cost linear in the number of samples, and it is proved that under a mean-shift alternative, the test always has greater relative efficiency than a previous linear-time kernel test, regardless of the choice of parameters for that test.
Squared-loss Mutual Information Regularization: A Novel Information-theoretic Approach to Semi-supervised Learning
SMIR is convex under mild conditions and thus improves the nonconvexity of mutual information regularization, and offers all of the following four abilities to semi-supervised algorithms: Analytical solution, out-of-sample/multi-class classification, and probabilistic output.
Large sample analysis of the median heuristic
In theory, this paper provides a convergence analysis that shows the asymptotic normality of the bandwidth chosen by the median heuristic in the setting of kernel two-sample test.
Bayesian Manifold Learning: The Locally Linear Latent Variable Model (LL-LVM)
- Mijung Park, Wittawat Jitkrittum, A. Qamar, Z. Szabó, Lars Buesing, M. Sahani
- Computer ScienceNIPS
- 24 October 2014
The LL-LVM encapsulates the local-geometry preserving intuitions that underlie non-probabilistic methods such as locally linear embedding (LLE) and makes it easy to evaluate the quality of hypothesised neighbourhood relationships, select the intrinsic dimensionality of the manifold, construct out-of-sample extensions and to combine the manifold model with additional probabilistic models that capture the structure of coordinates within the manifold.
An Adaptive Test of Independence with Analytic Kernel Embeddings
- Wittawat Jitkrittum, Z. Szabó, A. Gretton
- Computer ScienceInternational Conference on Machine Learning
- 15 October 2016
A new computationally efficient dependence measure, and an adaptive statistical test of independence, are proposed, which perform comparably to the state-of-the-art quadratic-time HSIC test, and outperform competing O( n) and O(n log n) tests.
Kernel-Based Just-In-Time Learning for Passing Expectation Propagation Messages
- Wittawat Jitkrittum, A. Gretton, Z. Szabó
- Computer ScienceConference on Uncertainty in Artificial…
- 9 March 2015
We propose an efficient nonparametric strategy for learning a message operator in expectation propagation (EP), which takes as input the set of incoming messages to a factor node, and produces an…
Learning Kernel Tests Without Data Splitting
- Jonas M. Kubler, Wittawat Jitkrittum, B. Scholkopf, Krikamol Muandet
- Computer ScienceNeural Information Processing Systems
- 1 June 2020
This work proposes an approach that enables learning the hyperparameters and testing on the full sample without data splitting, and can correctly calibrate the test in the presence of such dependency, and yield a test threshold in closed form.