Learning Generative Models using Denoising Density Estimators
This work addresses a fundamental problem in unsupervised machine learning for researchers and practitioners, offering a novel method that is incremental in improving upon existing generative techniques.
The paper tackles the challenge of learning probabilistic models for density estimation and sample generation by introducing a new generative model based on denoising density estimators (DDEs), which leverages neural networks to efficiently represent kernel density estimators and minimizes KL-divergence directly, resulting in substantial improvement in density estimation and competitive generative performance.
Learning probabilistic models that can estimate the density of a given set of samples, and generate samples from that density, is one of the fundamental challenges in unsupervised machine learning. We introduce a new generative model based on denoising density estimators (DDEs), which are scalar functions parameterized by neural networks, that are efficiently trained to represent kernel density estimators of the data. Leveraging DDEs, our main contribution is a novel technique to obtain generative models by minimizing the KL-divergence directly. We prove that our algorithm for obtaining generative models is guaranteed to converge to the correct solution. Our approach does not require specific network architecture as in normalizing flows, nor use ordinary differential equation solvers as in continuous normalizing flows. Experimental results demonstrate substantial improvement in density estimation and competitive performance in generative model training.