Proceedings Article•
An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale
TL;DR: The Vision Transformer (ViT) as discussed by the authors uses a pure transformer applied directly to sequences of image patches to perform very well on image classification tasks, achieving state-of-the-art results on ImageNet, CIFAR-100, VTAB, etc.
Abstract: While the Transformer architecture has become the de-facto standard for natural language processing tasks, its applications to computer vision remain limited. In vision, attention is either applied in conjunction with convolutional networks, or used to replace certain components of convolutional networks while keeping their overall structure in place. We show that this reliance on CNNs is not necessary and a pure transformer applied directly to sequences of image patches can perform very well on image classification tasks. When pre-trained on large amounts of data and transferred to multiple mid-sized or small image recognition benchmarks (ImageNet, CIFAR-100, VTAB, etc.), Vision Transformer (ViT) attains excellent results compared to state-of-the-art convolutional networks while requiring substantially fewer computational resources to train.
Citations
More filters
Posted Content•
TL;DR: Schlag and van Steenkiste as discussed by the authors used the resources usedfor the project were partially provided by Swiss National Supercomputing Centre (CSCS) project s1023.
Abstract: We thank Imanol Schlag and Sjoerd van Steenkiste for helpful discussions and suggestions on an earlier version of the manuscript. This research was partially funded by ERC Advanced grant no: 742870, project AlgoRNN, and by Swiss National Science Foundation grant no: 200021 192356, project NEUSYM. We are thankful for hardware donations from NVIDIA & IBM. The resources usedfor the project were partially provided by Swiss National Supercomputing Centre (CSCS) project s1023.
1 citations
Posted Content•
TL;DR: In this article, the authors design a training setup with several shortcut cues, named WCST-ML, where each cue is equally conducive to the visual recognition problem at hand, and observe that certain cues are preferred to others, solutions biased to the easy-to-learn cues tend to converge to relatively flat minima on the loss surface.
Abstract: Deep neural networks (DNNs) often rely on easy-to-learn discriminatory features, or cues, that are not necessarily essential to the problem at hand. For example, ducks in an image may be recognized based on their typical background scenery, such as lakes or streams. This phenomenon, also known as shortcut learning, is emerging as a key limitation of the current generation of machine learning models. In this work, we introduce a set of experiments to deepen our understanding of shortcut learning and its implications. We design a training setup with several shortcut cues, named WCST-ML, where each cue is equally conducive to the visual recognition problem at hand. Even under equal opportunities, we observe that (1) certain cues are preferred to others, (2) solutions biased to the easy-to-learn cues tend to converge to relatively flat minima on the loss surface, and (3) the solutions focusing on those preferred cues are far more abundant in the parameter space. We explain the abundance of certain cues via their Kolmogorov (descriptional) complexity: solutions corresponding to Kolmogorov-simple cues are abundant in the parameter space and are thus preferred by DNNs. Our studies are based on the synthetic dataset DSprites and the face dataset UTKFace. In our WCST-ML, we observe that the inborn bias of models leans toward simple cues, such as color and ethnicity. Our findings emphasize the importance of active human intervention to remove the inborn model biases that may cause negative societal impacts.
1 citations
Posted Content•
TL;DR: In this paper, the authors evaluate the adversarial robustness of the Vision Transformer (ViT) and MLP-Mixer architectures under several adversarial attack setups and benchmark them against the widely used CNNs.
Abstract: Convolutional Neural Networks (CNNs) have become the de facto gold standard in computer vision applications in the past years. Recently, however, new model architectures have been proposed challenging the status quo. The Vision Transformer (ViT) relies solely on attention modules, while the MLP-Mixer architecture substitutes the self-attention modules with Multi-Layer Perceptrons (MLPs). Despite their great success, CNNs have been widely known to be vulnerable to adversarial attacks, causing serious concerns for security-sensitive applications. Thus, it is critical for the community to know whether the newly proposed ViT and MLP-Mixer are also vulnerable to adversarial attacks. To this end, we empirically evaluate their adversarial robustness under several adversarial attack setups and benchmark them against the widely used CNNs. Overall, we find that the two architectures, especially ViT, are more robust than their CNN models. Using a toy example, we also provide empirical evidence that the lower adversarial robustness of CNNs can be partially attributed to their shift-invariant property. Our frequency analysis suggests that the most robust ViT architectures tend to rely more on low-frequency features compared with CNNs. Additionally, we have an intriguing finding that MLP-Mixer is extremely vulnerable to universal adversarial perturbations.
1 citations
Posted Content•
TL;DR: Wang et al. as discussed by the authors proposed Mask-guided Spectral-wise Transformer (MST) for hyperspectral image (HSI) reconstruction, which treats each spectral feature as a token and calculates self-attention along the spectral dimension.
Abstract: Hyperspectral image (HSI) reconstruction aims to recover the 3D spatial-spectral signal from a 2D measurement in the coded aperture snapshot spectral imaging (CASSI) system. The HSI representations are highly similar and correlated across the spectral dimension. Modeling the inter-spectra interactions is beneficial for HSI reconstruction. However, existing CNN-based methods show limitations in capturing spectral-wise similarity and long-range dependencies. Besides, the HSI information is modulated by a coded aperture (physical mask) in CASSI. Nonetheless, current algorithms have not fully explored the guidance effect of the mask for HSI restoration. In this paper, we propose a novel framework, Mask-guided Spectral-wise Transformer (MST), for HSI reconstruction. Specifically, we present a Spectral-wise Multi-head Self-Attention (S-MSA) that treats each spectral feature as a token and calculates self-attention along the spectral dimension. In addition, we customize a Mask-guided Mechanism (MM) that directs S-MSA to pay attention to spatial regions with high-fidelity spectral representations. Extensive experiments show that our MST significantly outperforms state-of-the-art (SOTA) methods on simulation and real HSI datasets while requiring dramatically cheaper computational and memory costs.
1 citations
Posted Content•
TL;DR: In this article, the authors proposed StarVQA, a novel underline space-time attention network for video quality assessment, which is trained using a vectorized regression loss by encoding the mean opinion score (MOS) to the probability vector and embedding a special vectorized label token as the learnable variable.
Abstract: The attention mechanism is blooming in computer vision nowadays. However, its application to video quality assessment (VQA) has not been reported. Evaluating the quality of in-the-wild videos is challenging due to the unknown of pristine reference and shooting distortion. This paper presents a novel \underline{s}pace-\underline{t}ime \underline{a}ttention network fo\underline{r} the \underline{VQA} problem, named StarVQA. StarVQA builds a Transformer by alternately concatenating the divided space-time attention. To adapt the Transformer architecture for training, StarVQA designs a vectorized regression loss by encoding the mean opinion score (MOS) to the probability vector and embedding a special vectorized label token as the learnable variable. To capture the long-range spatiotemporal dependencies of a video sequence, StarVQA encodes the space-time position information of each patch to the input of the Transformer. Various experiments are conducted on the de-facto in-the-wild video datasets, including LIVE-VQC, KoNViD-1k, LSVQ, and LSVQ-1080p. Experimental results demonstrate the superiority of the proposed StarVQA over the state-of-the-art. Code and model will be available at: this https URL.
1 citations
References
More filters
27 Jun 2016
TL;DR: In this article, the authors proposed a residual learning framework to ease the training of networks that are substantially deeper than those used previously, which won the 1st place on the ILSVRC 2015 classification task.
Abstract: Deeper neural networks are more difficult to train. We present a residual learning framework to ease the training of networks that are substantially deeper than those used previously. We explicitly reformulate the layers as learning residual functions with reference to the layer inputs, instead of learning unreferenced functions. We provide comprehensive empirical evidence showing that these residual networks are easier to optimize, and can gain accuracy from considerably increased depth. On the ImageNet dataset we evaluate residual nets with a depth of up to 152 layers—8× deeper than VGG nets [40] but still having lower complexity. An ensemble of these residual nets achieves 3.57% error on the ImageNet test set. This result won the 1st place on the ILSVRC 2015 classification task. We also present analysis on CIFAR-10 with 100 and 1000 layers. The depth of representations is of central importance for many visual recognition tasks. Solely due to our extremely deep representations, we obtain a 28% relative improvement on the COCO object detection dataset. Deep residual nets are foundations of our submissions to ILSVRC & COCO 2015 competitions1, where we also won the 1st places on the tasks of ImageNet detection, ImageNet localization, COCO detection, and COCO segmentation.
123,388 citations
Proceedings Article•
01 Jan 2015TL;DR: This work introduces Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments, and provides a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework.
Abstract: We introduce Adam, an algorithm for first-order gradient-based optimization of stochastic objective functions, based on adaptive estimates of lower-order moments. The method is straightforward to implement, is computationally efficient, has little memory requirements, is invariant to diagonal rescaling of the gradients, and is well suited for problems that are large in terms of data and/or parameters. The method is also appropriate for non-stationary objectives and problems with very noisy and/or sparse gradients. The hyper-parameters have intuitive interpretations and typically require little tuning. Some connections to related algorithms, on which Adam was inspired, are discussed. We also analyze the theoretical convergence properties of the algorithm and provide a regret bound on the convergence rate that is comparable to the best known results under the online convex optimization framework. Empirical results demonstrate that Adam works well in practice and compares favorably to other stochastic optimization methods. Finally, we discuss AdaMax, a variant of Adam based on the infinity norm.
111,197 citations
Proceedings Article•
03 Dec 2012TL;DR: The state-of-the-art performance of CNNs was achieved by Deep Convolutional Neural Networks (DCNNs) as discussed by the authors, which consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax.
Abstract: We trained a large, deep convolutional neural network to classify the 1.2 million high-resolution images in the ImageNet LSVRC-2010 contest into the 1000 different classes. On the test data, we achieved top-1 and top-5 error rates of 37.5% and 17.0% which is considerably better than the previous state-of-the-art. The neural network, which has 60 million parameters and 650,000 neurons, consists of five convolutional layers, some of which are followed by max-pooling layers, and three fully-connected layers with a final 1000-way softmax. To make training faster, we used non-saturating neurons and a very efficient GPU implementation of the convolution operation. To reduce overriding in the fully-connected layers we employed a recently-developed regularization method called "dropout" that proved to be very effective. We also entered a variant of this model in the ILSVRC-2012 competition and achieved a winning top-5 test error rate of 15.3%, compared to 26.2% achieved by the second-best entry.
73,978 citations
20 Jun 2009
TL;DR: A new database called “ImageNet” is introduced, a large-scale ontology of images built upon the backbone of the WordNet structure, much larger in scale and diversity and much more accurate than the current image datasets.
Abstract: The explosion of image data on the Internet has the potential to foster more sophisticated and robust models and algorithms to index, retrieve, organize and interact with images and multimedia data. But exactly how such data can be harnessed and organized remains a critical problem. We introduce here a new database called “ImageNet”, a large-scale ontology of images built upon the backbone of the WordNet structure. ImageNet aims to populate the majority of the 80,000 synsets of WordNet with an average of 500-1000 clean and full resolution images. This will result in tens of millions of annotated images organized by the semantic hierarchy of WordNet. This paper offers a detailed analysis of ImageNet in its current state: 12 subtrees with 5247 synsets and 3.2 million images in total. We show that ImageNet is much larger in scale and diversity and much more accurate than the current image datasets. Constructing such a large-scale database is a challenging task. We describe the data collection scheme with Amazon Mechanical Turk. Lastly, we illustrate the usefulness of ImageNet through three simple applications in object recognition, image classification and automatic object clustering. We hope that the scale, accuracy, diversity and hierarchical structure of ImageNet can offer unparalleled opportunities to researchers in the computer vision community and beyond.
49,639 citations
11 Oct 2018
TL;DR: BERT as mentioned in this paper pre-trains deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers, which can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks.
Abstract: We introduce a new language representation model called BERT, which stands for Bidirectional Encoder Representations from Transformers. Unlike recent language representation models (Peters et al., 2018a; Radford et al., 2018), BERT is designed to pre-train deep bidirectional representations from unlabeled text by jointly conditioning on both left and right context in all layers. As a result, the pre-trained BERT model can be fine-tuned with just one additional output layer to create state-of-the-art models for a wide range of tasks, such as question answering and language inference, without substantial task-specific architecture modifications. BERT is conceptually simple and empirically powerful. It obtains new state-of-the-art results on eleven natural language processing tasks, including pushing the GLUE score to 80.5 (7.7 point absolute improvement), MultiNLI accuracy to 86.7% (4.6% absolute improvement), SQuAD v1.1 question answering Test F1 to 93.2 (1.5 point absolute improvement) and SQuAD v2.0 Test F1 to 83.1 (5.1 point absolute improvement).
24,672 citations