Selecting Data Augmentation for Simulating Interventions

Ilse, Maximilian; Tomczak, Jakub M.; Forré, Patrick

doi:10.48550/arxiv.2005.01856

Cited by 5 publications

(7 citation statements)

References 21 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This assumption is not feasible if we do not have access to this model. Similar connections of causality and data augmentation have been made in (Ilse et al, 2020). It is also possible to train on interventional images x IV , i.e., generating a single image per sampled noise vector.…”

Section: Training An Invariant Classifiermentioning

confidence: 90%

Counterfactual Generative Networks

Sauer,

Geiger

2021

Preprint

View full text Add to dashboard Cite

Neural networks are prone to learning shortcuts -they often model simple correlations, ignoring more complex ones that potentially generalize better. Prior works on image classification show that instead of learning a connection to object shape, deep classifiers tend to exploit spurious correlations with low-level texture or the background for solving the classification task. In this work, we take a step towards more robust and interpretable classifiers that explicitly expose the task's causal structure. Building on current advances in deep generative modeling, we propose to decompose the image generation process into independent causal mechanisms that we train without direct supervision. By exploiting appropriate inductive biases, these mechanisms disentangle object shape, object texture, and background; hence, they allow for generating counterfactual images. We demonstrate the ability of our model to generate such images on MNIST and ImageNet. Further, we show that the counterfactual images can improve out-of-distribution robustness with a marginal drop in performance on the original classification task, despite being synthetic. Lastly, our generative model can be trained efficiently on a single GPU, exploiting common pre-trained models as inductive biases.

show abstract

Section: Training An Invariant Classifiermentioning

confidence: 90%

Counterfactual Generative Networks

Sauer,

Geiger

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…These methods essentially minimize the differences among domain specific risks. Moreover, data augmentation can also improve the generalizability of DNNs from the data perspective [12,26]. However, it requires prior knowledge on the differences between training and test domains, which are not be available in OOD prediction.…”

Section: An Effective Fix For Irmmentioning

confidence: 99%

Out-of-distribution Prediction with Invariant Risk Minimization: The Limitation and An Effective Fix

Guo¹,

Zhang²,

Líu³

et al. 2021

Preprint

View full text Add to dashboard Cite

This work considers the out-of-distribution (OOD) prediction problem where (1) the training data are from multiple domains and (2) the test domain is unseen in the training. DNNs fail in OOD prediction because they are prone to pick up spurious correlations. Recently, Invariant Risk Minimization (IRM) is proposed to address this issue. Its effectiveness has been demonstrated in the colored MNIST experiment. Nevertheless, we find that the performance of IRM can be dramatically degraded under strong Λ spuriousness -when the spurious correlation between the spurious features and the class label is strong due to the strong causal influence of their common cause, the domain label, on both of them (see Fig. 1). In this work, we try to answer the questions: why does IRM fail in the aforementioned setting? Why does IRM work for the original colored MNIST dataset? Then, we propose a simple and effective approach to fix the problem of IRM. We combine IRM with conditional distribution matching to avoid a specific type of spurious correlation under strong Λ spuriousness. Empirically, we design a series of semi synthetic datasetsthe colored MNIST plus, which exposes the problems of IRM and demonstrates the efficacy of the proposed method.

show abstract

“…. We measure the policy distance using Jensen-Shannon Divergence because it is a lower bound for the joint empirical risk across nonaugmented and augmented observations (Ilse et al, 2020;Raileanu et al, 2020). The policy distances are remarkably reduced when using augmentation in spite of delayed usage at (10, 25), (20,25) We analyze about two samples, which are trained on Bigfish environments, for verifying when generalization occurs.…”

Section: Augmentation During Rl Trainingmentioning

confidence: 99%

Efficient Scheduling of Data Augmentation for Deep Reinforcement Learning

Byungchan¹,

Ok²

2021

Preprint

View full text Add to dashboard Cite

Data augmentation technique from computer vision has been widely considered as a regularization method to improve data efficiency and generalization performance in vision-based reinforcement learning. We variate the timing of using augmentation, which is, in turn, critical depending on tasks to be solved in training and testing. According to our experiments on Open AI Procgen Benchmark, if the regularization imposed by augmentation is helpful only in testing, it is better to procrastinate the augmentation after training than to use it during training in terms of sample and computation complexity. We note that some of such augmentations can disturb the training process. Conversely, an augmentation providing regularization useful in training needs to be used during the whole training period to fully utilize its benefit in terms of not only generalization but also data efficiency. These phenomena suggest a useful timing control of data augmentation in reinforcement learning.

show abstract

Selecting Data Augmentation for Simulating Interventions

Cited by 5 publications

References 21 publications

Counterfactual Generative Networks

Counterfactual Generative Networks

Out-of-distribution Prediction with Invariant Risk Minimization: The Limitation and An Effective Fix

Efficient Scheduling of Data Augmentation for Deep Reinforcement Learning

Contact Info

Product

Resources

About