MintNet: Building Invertible Neural Networks with Masked Convolutions

NeurIPS 2019  ·  Yang Song, Chenlin Meng, Stefano Ermon ·

We propose a new way of constructing invertible neural networks by combining simple building blocks with a novel set of composition rules. This leads to a rich set of invertible architectures, including those similar to ResNets. Inversion is achieved with a locally convergent iterative procedure that is parallelizable and very fast in practice. Additionally, the determinant of the Jacobian can be computed analytically and efficiently, enabling their generative use as flow models. To demonstrate their flexibility, we show that our invertible neural networks are competitive with ResNets on MNIST and CIFAR-10 classification. When trained as generative models, our invertible networks achieve competitive likelihoods on MNIST, CIFAR-10 and ImageNet 32x32, with bits per dimension of 0.98, 3.32 and 4.06 respectively.

PDF Abstract NeurIPS 2019 PDF NeurIPS 2019 Abstract

Datasets


Results from the Paper


Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Image Generation CIFAR-10 MintNet bits/dimension 3.32 # 54
Image Generation ImageNet 32x32 MintNet bpd 4.06 # 23
Image Generation MNIST MintNet bits/dimension 0.98 # 4

Methods


No methods listed for this paper. Add relevant methods here