self training with noisy student improves imagenet classification

Noisy student-teacher training for robust keyword spotting, Unsupervised Self-training Algorithm Based on Deep Learning for Optical The proposed use of distillation to only handle easy instances allows for a more aggressive trade-off in the student size, thereby reducing the amortized cost of inference and achieving better accuracy than standard distillation. However, during the learning of the student, we inject noise such as dropout, stochastic depth and data augmentation via RandAugment to the student so that the student generalizes better than the teacher. As a comparison, our method only requires 300M unlabeled images, which is perhaps more easy to collect. Our study shows that using unlabeled data improves accuracy and general robustness. Overall, EfficientNets with Noisy Student provide a much better tradeoff between model size and accuracy when compared with prior works. We iterate this process by putting back the student as the teacher. Here we study if it is possible to improve performance on small models by using a larger teacher model, since small models are useful when there are constraints for model size and latency in real-world applications. A. Krizhevsky, I. Sutskever, and G. E. Hinton, Temporal ensembling for semi-supervised learning, Pseudo-label: the simple and efficient semi-supervised learning method for deep neural networks, Workshop on Challenges in Representation Learning, ICML, Certainty-driven consistency loss for semi-supervised learning, C. Liu, B. Zoph, M. Neumann, J. Shlens, W. Hua, L. Li, L. Fei-Fei, A. Yuille, J. Huang, and K. Murphy, R. G. Lopes, D. Yin, B. Poole, J. Gilmer, and E. D. Cubuk, Improving robustness without sacrificing accuracy with patch gaussian augmentation, Y. Luo, J. Zhu, M. Li, Y. Ren, and B. Zhang, Smooth neighbors on teacher graphs for semi-supervised learning, L. Maale, C. K. Snderby, S. K. Snderby, and O. Winther, A. Madry, A. Makelov, L. Schmidt, D. Tsipras, and A. Vladu, Towards deep learning models resistant to adversarial attacks, D. Mahajan, R. Girshick, V. Ramanathan, K. He, M. Paluri, Y. Li, A. Bharambe, and L. van der Maaten, Exploring the limits of weakly supervised pretraining, T. Miyato, S. Maeda, S. Ishii, and M. Koyama, Virtual adversarial training: a regularization method for supervised and semi-supervised learning, IEEE transactions on pattern analysis and machine intelligence, A. Najafi, S. Maeda, M. Koyama, and T. Miyato, Robustness to adversarial perturbations in learning from incomplete data, J. Ngiam, D. Peng, V. Vasudevan, S. Kornblith, Q. V. Le, and R. Pang, Robustness properties of facebooks resnext wsl models, Adversarial dropout for supervised and semi-supervised learning, Lessons from building acoustic models with a million hours of speech, IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), S. Qiao, W. Shen, Z. Zhang, B. Wang, and A. Yuille, Deep co-training for semi-supervised image recognition, I. Radosavovic, P. Dollr, R. Girshick, G. Gkioxari, and K. He, Data distillation: towards omni-supervised learning, A. Rasmus, M. Berglund, M. Honkala, H. Valpola, and T. Raiko, Semi-supervised learning with ladder networks, E. Real, A. Aggarwal, Y. Huang, and Q. V. Le, Proceedings of the AAAI Conference on Artificial Intelligence, B. Recht, R. Roelofs, L. Schmidt, and V. Shankar. Significantly, after using the masks generated by student-SN, the classification performance improved by 0.9 of AC, 0.7 of SE, and 0.9 of AUC. Here we study how to effectively use out-of-domain data. A common workaround is to use entropy minimization or ramp up the consistency loss. [68, 24, 55, 22]. We investigate the importance of noising in two scenarios with different amounts of unlabeled data and different teacher model accuracies. Their framework is highly optimized for videos, e.g., prediction on which frame to use in a video, which is not as general as our work. In our experiments, we use dropout[63], stochastic depth[29], data augmentation[14] to noise the student. Noisy Student self-training is an effective way to leverage unlabelled datasets and improving accuracy by adding noise to the student model while training so it learns beyond the teacher's knowledge. Chum, Label propagation for deep semi-supervised learning, D. P. Kingma, S. Mohamed, D. J. Rezende, and M. Welling, Semi-supervised learning with deep generative models, Semi-supervised classification with graph convolutional networks. We use our best model Noisy Student with EfficientNet-L2 to teach student models with sizes ranging from EfficientNet-B0 to EfficientNet-B7. On ImageNet, we first train an EfficientNet model on labeled images and use it as a teacher to generate pseudo labels for 300M unlabeled images. To date (2020) we will introduce "Noisy Student Training", which is a state-of-the-art model.The idea is to extend self-training and Distillation, a paper that shows that by adding three noises and distilling multiple times, the student model will have better generalization performance than the teacher model. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (pp. On, International journal of molecular sciences. Self-Training With Noisy Student Improves ImageNet Classification Abstract: We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. labels, the teacher is not noised so that the pseudo labels are as good as Code for Noisy Student Training. You can also use the colab script noisystudent_svhn.ipynb to try the method on free Colab GPUs. Our largest model, EfficientNet-L2, needs to be trained for 3.5 days on a Cloud TPU v3 Pod, which has 2048 cores. Our work is based on self-training (e.g.,[59, 79, 56]). We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. On ImageNet-C, it reduces mean corruption error (mCE) from 45.7 to 31.2. Figure 1(a) shows example images from ImageNet-A and the predictions of our models. We then train a larger EfficientNet as a student model on the combination of labeled and pseudo labeled images. (using extra training data). This invariance constraint reduces the degrees of freedom in the model. Do imagenet classifiers generalize to imagenet? We apply dropout to the final classification layer with a dropout rate of 0.5. A new scaling method is proposed that uniformly scales all dimensions of depth/width/resolution using a simple yet highly effective compound coefficient and is demonstrated the effectiveness of this method on scaling up MobileNets and ResNet. Self-training with Noisy Student improves ImageNet classification. By clicking accept or continuing to use the site, you agree to the terms outlined in our. We use EfficientNet-B0 as both the teacher model and the student model and compare using Noisy Student with soft pseudo labels and hard pseudo labels. During the generation of the pseudo labels, the teacher is not noised so that the pseudo labels are as accurate as possible. Noisy Student Training is based on the self-training framework and trained with 4 simple steps: For ImageNet checkpoints trained by Noisy Student Training, please refer to the EfficientNet github. 3.5B weakly labeled Instagram images. 10687-10698). 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition. We sample 1.3M images in confidence intervals. Hence, a question that naturally arises is why the student can outperform the teacher with soft pseudo labels. If nothing happens, download GitHub Desktop and try again. augmentation, dropout, stochastic depth to the student so that the noised In other words, using Noisy Student makes a much larger impact to the accuracy than changing the architecture. Noisy Student Training is a semi-supervised training method which achieves 88.4% top-1 accuracy on ImageNet Finally, we iterate the process by putting back the student as a teacher to generate new pseudo labels and train a new student. Models are available at https://github.com/tensorflow/tpu/tree/master/models/official/efficientnet. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to . Noisy Student (B7) means to use EfficientNet-B7 for both the student and the teacher. E. Arazo, D. Ortego, P. Albert, N. E. OConnor, and K. McGuinness, Pseudo-labeling and confirmation bias in deep semi-supervised learning, B. Athiwaratkun, M. Finzi, P. Izmailov, and A. G. Wilson, There are many consistent explanations of unlabeled data: why you should average, International Conference on Learning Representations, Advances in Neural Information Processing Systems, D. Berthelot, N. Carlini, I. Goodfellow, N. Papernot, A. Oliver, and C. Raffel, MixMatch: a holistic approach to semi-supervised learning, Combining labeled and unlabeled data with co-training, C. Bucilu, R. Caruana, and A. Niculescu-Mizil, Proceedings of the 12th ACM SIGKDD international conference on Knowledge discovery and data mining, Y. Carmon, A. Raghunathan, L. Schmidt, P. Liang, and J. C. Duchi, Unlabeled data improves adversarial robustness, Semi-supervised learning (chapelle, o. et al., eds. On ImageNet, we first train an EfficientNet model on labeled images and use it as a teacher to generate pseudo labels for 300M unlabeled images. But training robust supervised learning models is requires this step. To achieve this result, we first train an EfficientNet model on labeled ImageNet images and use it as a teacher to generate pseudo labels on 300M unlabeled images. [76] also proposed to first only train on unlabeled images and then finetune their model on labeled images as the final stage. Our main results are shown in Table1. For more information about the large architectures, please refer to Table7 in Appendix A.1. We found that self-training is a simple and effective algorithm to leverage unlabeled data at scale. mFR (mean flip rate) is the weighted average of flip probability on different perturbations, with AlexNets flip probability as a baseline. Instructions on running prediction on unlabeled data, filtering and balancing data and training using the stored predictions. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2. Next, with the EfficientNet-L0 as the teacher, we trained a student model EfficientNet-L1, a wider model than L0. task. However, during the learning of the student, we inject noise such as dropout, stochastic depth and data augmentation via RandAugment to the student so that the student generalizes better than the teacher. Specifically, as all classes in ImageNet have a similar number of labeled images, we also need to balance the number of unlabeled images for each class. The swing in the picture is barely recognizable by human while the Noisy Student model still makes the correct prediction. (or is it just me), Smithsonian Privacy Stochastic depth is proposed, a training procedure that enables the seemingly contradictory setup to train short networks and use deep networks at test time and reduces training time substantially and improves the test error significantly on almost all data sets that were used for evaluation. Self-training with Noisy Student improves ImageNet classification. First, we run an EfficientNet-B0 trained on ImageNet[69]. Le, and J. Shlens, Using videos to evaluate image model robustness, Deep residual learning for image recognition, Benchmarking neural network robustness to common corruptions and perturbations, D. Hendrycks, K. Zhao, S. Basart, J. Steinhardt, and D. Song, Distilling the knowledge in a neural network, G. Huang, Z. Liu, L. Van Der Maaten, and K. Q. Weinberger, G. Huang, Y. A tag already exists with the provided branch name. 2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). This is an important difference between our work and prior works on teacher-student framework whose main goal is model compression. Selected images from robustness benchmarks ImageNet-A, C and P. Test images from ImageNet-C underwent artificial transformations (also known as common corruptions) that cannot be found on the ImageNet training set. [50] used knowledge distillation on unlabeled data to teach a small student model for speech recognition. This way, we can isolate the influence of noising on unlabeled images from the influence of preventing overfitting for labeled images. Hence the total number of images that we use for training a student model is 130M (with some duplicated images). Use Git or checkout with SVN using the web URL. We use the same architecture for the teacher and the student and do not perform iterative training. In Noisy Student, we combine these two steps into one because it simplifies the algorithm and leads to better performance in our preliminary experiments. For a small student model, using our best model Noisy Student (EfficientNet-L2) as the teacher model leads to more improvements than using the same model as the teacher, which shows that it is helpful to push the performance with our method when small models are needed for deployment. Noisy Student Training achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. Especially unlabeled images are plentiful and can be collected with ease. Do better imagenet models transfer better? During the generation of the pseudo labels, the teacher is not noised so that the pseudo labels are as accurate as possible. The main difference between our method and knowledge distillation is that knowledge distillation does not consider unlabeled data and does not aim to improve the student model. This is why "Self-training with Noisy Student improves ImageNet classification" written by Qizhe Xie et al makes me very happy. mCE (mean corruption error) is the weighted average of error rate on different corruptions, with AlexNets error rate as a baseline. Self-training with noisy student improves imagenet classification, in: Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 10687-10698, (2020 . We also study the effects of using different amounts of unlabeled data. They did not show significant improvements in terms of robustness on ImageNet-A, C and P as we did. For classes where we have too many images, we take the images with the highest confidence. We then train a larger EfficientNet as a student model on the combination of labeled and pseudo labeled images. Are you sure you want to create this branch? In contrast, changing architectures or training with weakly labeled data give modest gains in accuracy from 4.7% to 16.6%. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2. When the student model is deliberately noised it is actually trained to be consistent to the more powerful teacher model that is not noised when it generates pseudo labels. During the generation of the pseudo We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. We vary the model size from EfficientNet-B0 to EfficientNet-B7[69] and use the same model as both the teacher and the student. Finally, in the above, we say that the pseudo labels can be soft or hard. For instance, on ImageNet-1k, Layer Grafted Pre-training yields 65.5% Top-1 accuracy in terms of 1% few-shot learning with ViT-B/16, which improves MIM and CL baselines by 14.4% and 2.1% with no bells and whistles. In the above experiments, iterative training was used to optimize the accuracy of EfficientNet-L2 but here we skip it as it is difficult to use iterative training for many experiments. all 12, Image Classification Their main goal is to find a small and fast model for deployment. Proceedings of the eleventh annual conference on Computational learning theory, Proceedings of the IEEE conference on computer vision and pattern recognition, Empirical Methods in Natural Language Processing (EMNLP), Imagenet classification with deep convolutional neural networks, Domain adaptive transfer learning with specialist models, Thirty-Second AAAI Conference on Artificial Intelligence, Regularized evolution for image classifier architecture search, Inception-v4, inception-resnet and the impact of residual connections on learning. EfficientNet with Noisy Student produces correct top-1 predictions (shown in. A tag already exists with the provided branch name. Noisy Student Training extends the idea of self-training and distillation with the use of equal-or-larger student models and noise added to the student during learning. Lastly, we will show the results of benchmarking our model on robustness datasets such as ImageNet-A, C and P and adversarial robustness. Noisy Student can still improve the accuracy to 1.6%. Many Git commands accept both tag and branch names, so creating this branch may cause unexpected behavior. While removing noise leads to a much lower training loss for labeled images, we observe that, for unlabeled images, removing noise leads to a smaller drop in training loss. It can be seen that masks are useful in improving classification performance. We present a simple self-training method that achieves 87.4 Self-training with Noisy Student improves ImageNet classification. Lastly, we trained another EfficientNet-L2 student by using the EfficientNet-L2 model as the teacher. Ranked #14 on "Self-training with Noisy Student improves ImageNet classification" pytorch implementation. Flip probability is the probability that the model changes top-1 prediction for different perturbations. The most interesting image is shown on the right of the first row. Computer Science - Computer Vision and Pattern Recognition. Noisy Student Training is a semi-supervised training method which achieves 88.4% top-1 accuracy on ImageNet and surprising gains on robustness and adversarial benchmarks. The top-1 accuracy is simply the average top-1 accuracy for all corruptions and all severity degrees. During the learning of the student, we inject noise such as dropout, stochastic depth, and data augmentation via RandAugment to the student so that the student generalizes better than the teacher. In particular, we first perform normal training with a smaller resolution for 350 epochs. To achieve this result, we first train an EfficientNet model on labeled Noisy Student Training is based on the self-training framework and trained with 4 simple steps: Train a classifier on labeled data (teacher). The paradigm of pre-training on large supervised datasets and fine-tuning the weights on the target task is revisited, and a simple recipe that is called Big Transfer (BiT) is created, which achieves strong performance on over 20 datasets. We then train a larger EfficientNet as a student model on the combination of labeled and pseudo labeled images. First, a teacher model is trained in a supervised fashion. We then perform data filtering and balancing on this corpus. When dropout and stochastic depth are used, the teacher model behaves like an ensemble of models (when it generates the pseudo labels, dropout is not used), whereas the student behaves like a single model. Finally, we iterate the algorithm a few times by treating the student as a teacher to generate new pseudo labels and train a new student. Self-Training With Noisy Student Improves ImageNet Classification Abstract: We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. We present Noisy Student Training, a semi-supervised learning approach that works well even when labeled data is abundant. This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository. Authors: Qizhe Xie, Minh-Thang Luong, Eduard Hovy, Quoc V. Le Description: We present a simple self-training method that achieves 88.4% top-1 accuracy on ImageNet, which is 2.0% better than the state-of-the-art model that requires 3.5B weakly labeled Instagram images. As shown in Figure 3, Noisy Student leads to approximately 10% improvement in accuracy even though the model is not optimized for adversarial robustness. combination of labeled and pseudo labeled images. The accuracy is improved by about 10% in most settings. Amongst other components, Noisy Student implements Self-Training in the context of Semi-Supervised Learning. The ADS is operated by the Smithsonian Astrophysical Observatory under NASA Cooperative Learn more. Afterward, we further increased the student model size to EfficientNet-L2, with the EfficientNet-L1 as the teacher. This paper proposes to search for an architectural building block on a small dataset and then transfer the block to a larger dataset and introduces a new regularization technique called ScheduledDropPath that significantly improves generalization in the NASNet models. Hence, EfficientNet-L0 has around the same training speed with EfficientNet-B7 but more parameters that give it a larger capacity. Self-training was previously used to improve ResNet-50 from 76.4% to 81.2% top-1 accuracy[76] which is still far from the state-of-the-art accuracy. A self-training method that better adapt to the popular two stage training pattern for multi-label text classification under a semi-supervised scenario by continuously finetuning the semantic space toward increasing high-confidence predictions, intending to further promote the performance on target tasks. For this purpose, we use a much larger corpus of unlabeled images, where some images may not belong to any category in ImageNet. The architectures for the student and teacher models can be the same or different. The abundance of data on the internet is vast. On ImageNet, we first train an EfficientNet model on labeled images and use it as a teacher to generate pseudo labels for 300M unlabeled images. on ImageNet, which is 1.0 Self-Training achieved the state-of-the-art in ImageNet classification within the framework of Noisy Student [1]. We use EfficientNet-B4 as both the teacher and the student. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2.Noisy Student Training extends the idea of self-training and distillation with the use of equal-or-larger student models and noise added to the student during learning. This work introduces two challenging datasets that reliably cause machine learning model performance to substantially degrade and curates an adversarial out-of-distribution detection dataset called IMAGENET-O, which is the first out- of-dist distribution detection dataset created for ImageNet models. Self-training We first improved the accuracy of EfficientNet-B7 using EfficientNet-B7 as both the teacher and the student. Code for Noisy Student Training. On robustness test sets, it improves ImageNet-A top-1 accuracy from 61.0% to 83.7%, reduces ImageNet-C mean corruption error from 45.7 to 28.3, and reduces ImageNet-P mean flip rate from 27.8 to 12.2. Self-Training With Noisy Student Improves ImageNet Classification. For this purpose, we use the recently developed EfficientNet architectures[69] because they have a larger capacity than ResNet architectures[23]. We use the standard augmentation instead of RandAugment in this experiment. Our experiments showed that our model significantly improves accuracy on ImageNet-A, C and P without the need for deliberate data augmentation. As shown in Figure 1, Noisy Student leads to a consistent improvement of around 0.8% for all model sizes. Noisy Students performance improves with more unlabeled data. The method, named self-training with Noisy Student, also benefits from the large capacity of EfficientNet family. https://arxiv.org/abs/1911.04252, Accompanying notebook and sources to "A Guide to Pseudolabelling: How to get a Kaggle medal with only one model" (Dec. 2020 PyData Boston-Cambridge Keynote), Deep learning has shown remarkable successes in image recognition in recent years[35, 66, 62, 23, 69]. possible. Edit social preview. ImageNet-A test set[25] consists of difficult images that cause significant drops in accuracy to state-of-the-art models. Self-Training Noisy Student " " Self-Training . A. Alemi, Thirty-First AAAI Conference on Artificial Intelligence, C. Szegedy, W. Liu, Y. Jia, P. Sermanet, S. Reed, D. Anguelov, D. Erhan, V. Vanhoucke, and A. Rabinovich, C. Szegedy, V. Vanhoucke, S. Ioffe, J. Shlens, and Z. Wojna, Rethinking the inception architecture for computer vision, C. Szegedy, W. Zaremba, I. Sutskever, J. Bruna, D. Erhan, I. Goodfellow, and R. Fergus, EfficientNet: rethinking model scaling for convolutional neural networks, Mean teachers are better role models: weight-averaged consistency targets improve semi-supervised deep learning results, H. Touvron, A. Vedaldi, M. Douze, and H. Jgou, Fixing the train-test resolution discrepancy, V. Verma, A. Lamb, J. Kannala, Y. Bengio, and D. Lopez-Paz, Proceedings of the Twenty-Eighth International Joint Conference on Artificial Intelligence (IJCAI-19), J. Weston, F. Ratle, H. Mobahi, and R. Collobert, Deep learning via semi-supervised embedding, Q. Xie, Z. Dai, E. Hovy, M. Luong, and Q. V. Le, Unsupervised data augmentation for consistency training, S. Xie, R. Girshick, P. Dollr, Z. Tu, and K. He, Aggregated residual transformations for deep neural networks, I. Callback to apply noisy student self-training (a semi-supervised learning approach) based on: Xie, Q., Luong, M. T., Hovy, E., & Le, Q. V. (2020). On . Qizhe Xie, Eduard Hovy, Minh-Thang Luong, Quoc V. Le. This is probably because it is harder to overfit the large unlabeled dataset. EfficientNet-L0 is wider and deeper than EfficientNet-B7 but uses a lower resolution, which gives it more parameters to fit a large number of unlabeled images with similar training speed. Train a larger classifier on the combined set, adding noise (noisy student). The best model in our experiments is a result of iterative training of teacher and student by putting back the student as the new teacher to generate new pseudo labels. This model investigates a new method. In all previous experiments, the students capacity is as large as or larger than the capacity of the teacher model. The main use case of knowledge distillation is model compression by making the student model smaller. The main difference between our work and prior works is that we identify the importance of noise, and aggressively inject noise to make the student better. We present Noisy Student Training, a semi-supervised learning approach that works well even when labeled data is abundant. Please refer to [24] for details about mCE and AlexNets error rate. To achieve strong results on ImageNet, the student model also needs to be large, typically larger than common vision models, so that it can leverage a large number of unlabeled images. We use EfficientNets[69] as our baseline models because they provide better capacity for more data. When data augmentation noise is used, the student must ensure that a translated image, for example, should have the same category with a non-translated image. We duplicate images in classes where there are not enough images. CLIP (Contrastive Language-Image Pre-training) builds on a large body of work on zero-shot transfer, natural language supervision, and multimodal learning.The idea of zero-data learning dates back over a decade [^reference-8] but until recently was mostly studied in computer vision as a way of generalizing to unseen object categories. We then select images that have confidence of the label higher than 0.3. We use the labeled images to train a teacher model using the standard cross entropy loss. In typical self-training with the teacher-student framework, noise injection to the student is not used by default, or the role of noise is not fully understood or justified. . We iterate this process by putting back the student as the teacher. To intuitively understand the significant improvements on the three robustness benchmarks, we show several images in Figure2 where the predictions of the standard model are incorrect and the predictions of the Noisy Student model are correct. We then use the teacher model to generate pseudo labels on unlabeled images. Noisy Student Training is based on the self-training framework and trained with 4-simple steps: This commit does not belong to any branch on this repository, and may belong to a fork outside of the repository. For unlabeled images, we set the batch size to be three times the batch size of labeled images for large models, including EfficientNet-B7, L0, L1 and L2. (2) With out-of-domain unlabeled images, hard pseudo labels can hurt the performance while soft pseudo labels leads to robust performance. Work fast with our official CLI. . Noisy StudentImageNetEfficientNet-L2state-of-the-art. Lastly, we follow the idea of compound scaling[69] and scale all dimensions to obtain EfficientNet-L2. Agreement NNX16AC86A, Is ADS down? Note that these adversarial robustness results are not directly comparable to prior works since we use a large input resolution of 800x800 and adversarial vulnerability can scale with the input dimension[17, 20, 19, 61].