Least-squares two-sample test

By Masashi Sugiyama, Taiji Suzuki, Yuta Itoh, Takafumi Kanamori, Manabu Kimura. 2011

45 citations recorded by the source corpus.

Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Statistics and Probability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computational Mechanics

1. Arthur Gretton
2. Masashi Sugiyama
3. Kenji Fukumizu
4. Bharath K. Sriperumbudur
5. Aaditya Ramdas
6. Hsiao-Yu Fish Tung
7. Alex Smola
8. Makoto Yamada

In Computer Vision and Pattern Recognition

1. Arthur Gretton
2. Kenji Fukumizu
3. Bharath K. Sriperumbudur
4. Masashi Sugiyama
5. Dino Sejdinovic
6. Bernhard Schölkopf
7. Danica J. Sutherland
8. Makoto Yamada

In Statistics and Probability

1. Arthur Gretton
2. Bharath K. Sriperumbudur
3. Kenji Fukumizu
4. Dino Sejdinovic
5. Bernhard Schölkopf
6. Kacper P. Chwialkowski
7. Krikamol Muandet
8. Heiko Strathmann

Citations in the corpus

• A Wild Bootstrap for Degenerate Kernel Tests (2014, methodology): “Statistical tests based on distribution embeddings into reproducing kernel Hilbert spaces have been applied in many contexts, including two sample testing [19, 15, 31], tests of independence [17, 32, 4], tests of conditional independence [14, 32], and tests for higher order (Lancaster) interactions ”
• Density-ratio matching under the Bregman divergence: a unified framework of density-ratio estimation (2012, unknown): “2009), two-sample test (Keziou and Leoni-Aubin 2005; Sugiyama et al. 2011a) change detection in time series (Kawahara and Sugiyama 2009), conditional density estimation (Sugiyama et al.”
• Optimal kernel choice for large-scale two-sample tests (2012, methodology): “A number of recent approaches to two-sample testing have made use of mappings of the distributions to a reproducing kernel Hilbert space (RKHS); or have sought out RKHS functions with large amplitude where the probability mass of p and q differs most [8, 10, 15, 17, 7].”
• Generative Models and Model Criticism via Optimized Maximum Mean Discrepancy (2016, methodology): “4 One can also consider maximizing criteria based on cross validation (Sugiyama et al., 2011; Gretton et al., 2012b; Strathmann, 2012).”
• Change-Point Detection in Time-Series Data by Relative Density-Ratio Estimation (2012, methodology): “…g ( Y ), an approximator of the PE divergence can be constructed as c b b This approximator is derived from the following expression of the PE divergence (Sugiyama et al., 2010, 2011b): The first two terms of (9) are actually the negative uLSIF optimization objective without regularization.”