Density Ratio Estimation

Density ratio estimation. The estimated density ratio function can be used in many applications such as anomaly detection, change-point detection, covariate shift adaptation. The implemented methods are uLSIF (Hido et al. (2011) ), RuLSIF (Yamada et al. (2011) ), and KLIEP (Sugiyama et al. (2007) ).

title: "An R Package for Density Ratio Estimation" author: Koji MAKIYAMA (@hoxo_m) output: html_document: keep_md: true md_document: variant: gfm

Density ratio estimation is described as follows: for given two data samples x and y from unknown distributions p(x) and q(y) respectively, estimate w(x) = p(x) / q(x), where x and y are d-dimensional real numbers.

The estimated density ratio function w(x) can be used in many applications such as anomaly detection [1] and covariate shift adaptation [2]. Other useful applications about density ratio estimation were summarized by Sugiyama et al. (2012) [3].

The package densratio provides a function densratio(). The function outputs an object that has a function to estimate density ratio.

For example,

set.seed(3)
x <- rnorm(200, mean = 1, sd = 1/8)
y <- rnorm(200, mean = 1, sd = 1/2)
 
library(densratio)
result <- densratio(x, y)

The function densratio() estimates the density ratio of p(x) to q(y), w(x) = p(x)/q(y) = Norm(1, 1/8) / Norm(1, 1/2), and provides a function to compute estimated density ratio. The result object has a function compute_density_ratio() that can compute the estimated density ratio w_hat(x) for any d-dimensional input x (now d=1).

new_x <- seq(0, 2, by = 0.06)
w_hat <- result$compute_density_ratio(new_x)
 
plot(new_x, w_hat, pch=19)

In this case, the true density ratio w(x) = p(x)/q(y) = Norm(1, 1/8) / Norm(1, 1/2) can be computed precisely. So we can compare w(x) with the estimated density ratio w_hat(x).

true_density_ratio <- function(x) dnorm(x, 1, 1/8) / dnorm(x, 1, 1/2)
 
plot(true_density_ratio, xlim=c(-1, 3), lwd=2, col="red", xlab = "x", ylab = "Density Ratio")
plot(result$compute_density_ratio, xlim=c(-1, 3), lwd=2, col="green", add=TRUE)
legend("topright", legend=c(expression(w(x)), expression(hat(w)(x))), col=2:3, lty=1, lwd=2, pch=NA)

2. Installation

You can install the densratio package from CRAN.

install.packages("densratio")

You can also install the package from GitHub.

install.packages("devtools") # if you have not installed "devtools" package
devtools::install_github("hoxo-m/densratio")

The source code for densratio package is available on GitHub at

• https://github.com/hoxo-m/densratio.

3. Details

densratio() has method argument that you can pass "uLSIF" or "KLIEP".

• uLSIF (unconstrained Least-Squares Importance Fitting) is the default method. This algorithm estimates density ratio by minimizing the squared loss. You can find more information in Hido et al. (2011) [1].

• KLIEP (Kullback-Leibler Importance Estimation Procedure) is the another method. This algorithm estimates density ratio by minimizing Kullback-Leibler divergence. You can find more information in Sugiyama et al. (2007) [2].

There is a vignette for the package. For more detail, read it.

vignette("densratio")

You can also find it on CRAN.

• An R Package for Density Ratio Estimation

4. Related work

We have also developed a Python package for density ratio estimation.

• https://github.com/hoxo-m/densratio_py

The package is available on PyPI (Python Package Index).

• https://pypi.python.org/pypi/densratio

5. References

[1] Hido, S., Tsuboi, Y., Kashima, H., Sugiyama, M., & Kanamori, T. Statistical outlier detection using direct density ratio estimation. Knowledge and Information Systems 2011.

[2] Sugiyama, M., Nakajima, S., Kashima, H., von Bünau, P. & Kawanabe, M. Direct importance estimation with model selection and its application to covariate shift adaptation. NIPS 2007.

[3] Sugiyama, M., Suzuki, T. & Kanamori, T. Density Ratio Estimation in Machine Learning. Cambridge University Press 2012.