Enforcing such independencies requires nuisances to be observed during training.
Learning models that gracefully handle distribution shifts is central to research on domain generalization, robust optimization, and fairness.
For problems where global invertibility is necessary, such as applying normalizing flows on OOD data, we show the importance of designing stable INN building blocks.
Deep learning has triggered the current rise of artificial intelligence and is the workhorse of today's machine intelligence.
Training neural ODEs on large datasets has not been tractable due to the necessity of allowing the adaptive numerical ODE solver to refine its step size to very small values.
Ranked #1 on Image Generation on ImageNet64x64
In this setting, the standard class probabilities can be easily computed as well as unnormalized values of p(x) and p(x|y).
Our BCOP parameterization allows us to train large convolutional networks with provable Lipschitz bounds.
Guarantees in deep learning are hard to achieve due to the interplay of flexible modeling schemes and complex tasks.
We consider the problem of learning representations that achieve group and subgroup fairness with respect to multiple sensitive attributes.
Flow-based generative models parameterize probability distributions through an invertible transformation and can be trained by maximum likelihood.
Ranked #2 on Image Generation on MNIST
Class-conditional generative models hold promise to overcome the shortcomings of their discriminative counterparts.
Excessive invariance is not limited to models trained to be robust to perturbation-based $\ell_p$-norm adversaries.
We show that standard ResNet architectures can be made invertible, allowing the same model to be used for classification, density estimation, and generation.
Ranked #5 on Image Generation on MNIST
An analysis of i-RevNets learned representations suggests an alternative explanation for the success of deep networks by a progressive contraction and linear separation with depth.
Filters in convolutional networks are typically parameterized in a pixel basis, that does not take prior knowledge about the visual world into account.
Multiscale hierarchical convolutional networks are structured deep convolutional networks where layers are indexed by progressively higher dimensional attributes, which are learned from training data.