Entropy-Informed Weighting Channel Normalizing Flow for Deep Generative Models
This work addresses a memory bottleneck in generative modeling for researchers and practitioners, offering an incremental improvement over existing multi-scale methods.
The paper tackled the memory inefficiency of Normalizing Flows in deep generative models by introducing a regularized, feature-dependent Shuffle operation into multi-scale architectures, resulting in state-of-the-art density estimation and competitive sample quality on datasets like CIFAR-10 and ImageNet with minimal computational overhead.
Normalizing Flows (NFs) are widely used in deep generative models for their exact likelihood estimation and efficient sampling. However, they require substantial memory since the latent space matches the input dimension. Multi-scale architectures address this by progressively reducing latent dimensions while preserving reversibility. Existing multi-scale architectures use simple, static channel-wise splitting, limiting expressiveness. To improve this, we introduce a regularized, feature-dependent $\mathtt{Shuffle}$ operation and integrate it into vanilla multi-scale architecture. This operation adaptively generates channel-wise weights and shuffles latent variables before splitting them. We observe that such operation guides the variables to evolve in the direction of entropy increase, hence we refer to NFs with the $\mathtt{Shuffle}$ operation as \emph{Entropy-Informed Weighting Channel Normalizing Flow} (EIW-Flow). Extensive experiments on CIFAR-10, CelebA, ImageNet, and LSUN demonstrate that EIW-Flow achieves state-of-the-art density estimation and competitive sample quality for deep generative modeling, with minimal computational overhead.