Learning Implicit Generative Models Using Differentiable Graph Tests

09/04/2017
by   Josip Djolonga, et al.
0

Recently, there has been a growing interest in the problem of learning rich implicit models - those from which we can sample, but can not evaluate their density. These models apply some parametric function, such as a deep network, to a base measure, and are learned end-to-end using stochastic optimization. One strategy of devising a loss function is through the statistics of two sample tests - if we can fool a statistical test, the learned distribution should be a good model of the true data. However, not all tests can easily fit into this framework, as they might not be differentiable with respect to the data points, and hence with respect to the parameters of the implicit model. Motivated by this problem, in this paper we show how two such classical tests, the Friedman-Rafsky and k-nearest neighbour tests, can be effectively smoothed using ideas from undirected graphical models - the matrix tree theorem and cardinality potentials. Moreover, as we show experimentally, smoothing can significantly increase the power of the test, which might of of independent interest. Finally, we apply our method to learn implicit models.

READ FULL TEXT

page 9

page 10

research
07/10/2018

Learning Implicit Generative Models by Teaching Explicit Ones

Implicit generative models are difficult to train as no explicit probabi...
research
02/24/2020

Testing Goodness of Fit of Conditional Density Models with Kernels

We propose two nonparametric statistical tests of goodness of fit for co...
research
03/02/2022

Model-agnostic out-of-distribution detection using combined statistical tests

We present simple methods for out-of-distribution detection using a trai...
research
06/16/2020

Goodness-of-Fit Test for Self-Exciting Processes

Recently there have been many research efforts in developing generative ...
research
10/11/2016

Learning in Implicit Generative Models

Generative adversarial networks (GANs) provide an algorithmic framework ...
research
07/14/2021

Feature Shift Detection: Localizing Which Features Have Shifted via Conditional Distribution Tests

While previous distribution shift detection approaches can identify if a...

Please sign up or login with your details

Forgot password? Click here to reset