Motivated by the fact that, in many communication schemes, the achievable transmission rate is determined by a conditional mutual information term, this paper focuses on neural-based estimators for this information-theoretic quantity. Our results are based on variational bounds for the KL-divergence ...
(2018) uses Gaussian processes in a Bayesian approach in which the model’s input is extended with latent variables with posterior distribution trained through a neural network. By contrast, the methodology of this article estimates conditional densities via conditional maps. A map-based density ...
On the other hand, it is shown (Watanabe, 2018b) that WAIC is an asymptotically unbiased estimator of the generalization loss, E[Gn]≅E[Wn], because the property of WAIC is derived from the convergence of the empirical process ξn(u)→ ξ(u) based on the conditional independence of ...
In this article, the author has presented a neural network-based conditional density estimator which is semi-non-parametric. In terms of time series modelling, his approach is more general than traditional GARCH models of asset return series because the shape of the conditional density depends on ...
with density p(∈ ϕ) with nonnegative support, and θ and ϕ are variation free. The baseline intensity, or baseline hazard, is given by (2.13)λ0=p(∈; ϕ)S(∈; ϕ), where S(∈; ϕ)=∫∈∞p(u; ϕ)du is the survivor function. The intensity func...
We introduce CaDET, an algorithm for parametric Conditional Density Estimation (CDE) based on decision trees and random forests. CaDET uses the empirical c
Different from the existing cGANs built upon the cross-entropy loss, our CCF-GAN benefits from the characteristic function (CF), which processes unique and universal correspondence to a random variable, even when the random variable does not possess probability densit...
Preconditioning an Artificial Neural Network Using Naive Bayes. Zaidi, N. A., Petitjean, F., & Webb, G. I. Proceedings of the 20th Pacific-Asia Conference on Advances in Knowledge Discovery and Data Mining, PAKDD 2016, pp. 341-353, 2016. ...
The final log density for the posterior distribution qϕ(z|x) after k steps builds up as: logqϕ(z|x)=logp(ϵ)−∑k=0K−1∑i=1Dlogσk,i(v) For the Autoregressive Neural Network (ARNN), we took a 2-layer MADE network, a Masked ARNN as introduced in Germain et al. ...
However, even if gross error occurs in response values, L1 estimator defined as an optimal solution of min w,b 1 m m |yi − f (xi; w, b)|, i=1 16 depresses influence of outliers. Square regularization term is often added to objective functions such as min w,b C m m |yi ...