{"title":"如何在小尺度数据集上训练视觉转换器?","authors":"Hanan Gani, Muzammal Naseer, Mohammad Yaqub","doi":"10.48550/arXiv.2210.07240","DOIUrl":null,"url":null,"abstract":"Vision Transformer (ViT), a radically different architecture than convolutional neural networks offers multiple advantages including design simplicity, robustness and state-of-the-art performance on many vision tasks. However, in contrast to convolutional neural networks, Vision Transformer lacks inherent inductive biases. Therefore, successful training of such models is mainly attributed to pre-training on large-scale datasets such as ImageNet with 1.2M or JFT with 300M images. This hinders the direct adaption of Vision Transformer for small-scale datasets. In this work, we show that self-supervised inductive biases can be learned directly from small-scale datasets and serve as an effective weight initialization scheme for fine-tuning. This allows to train these models without large-scale pre-training, changes to model architecture or loss functions. We present thorough experiments to successfully train monolithic and non-monolithic Vision Transformers on five small datasets including CIFAR10/100, CINIC10, SVHN, Tiny-ImageNet and two fine-grained datasets: Aircraft and Cars. Our approach consistently improves the performance of Vision Transformers while retaining their properties such as attention to salient regions and higher robustness. Our codes and pre-trained models are available at: https://github.com/hananshafi/vits-for-small-scale-datasets.","PeriodicalId":72437,"journal":{"name":"BMVC : proceedings of the British Machine Vision Conference. British Machine Vision Conference","volume":"20 1","pages":"731"},"PeriodicalIF":0.0000,"publicationDate":"2022-10-13","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"15","resultStr":"{\"title\":\"How to Train Vision Transformer on Small-scale Datasets?\",\"authors\":\"Hanan Gani, Muzammal Naseer, Mohammad Yaqub\",\"doi\":\"10.48550/arXiv.2210.07240\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"Vision Transformer (ViT), a radically different architecture than convolutional neural networks offers multiple advantages including design simplicity, robustness and state-of-the-art performance on many vision tasks. However, in contrast to convolutional neural networks, Vision Transformer lacks inherent inductive biases. Therefore, successful training of such models is mainly attributed to pre-training on large-scale datasets such as ImageNet with 1.2M or JFT with 300M images. This hinders the direct adaption of Vision Transformer for small-scale datasets. In this work, we show that self-supervised inductive biases can be learned directly from small-scale datasets and serve as an effective weight initialization scheme for fine-tuning. This allows to train these models without large-scale pre-training, changes to model architecture or loss functions. We present thorough experiments to successfully train monolithic and non-monolithic Vision Transformers on five small datasets including CIFAR10/100, CINIC10, SVHN, Tiny-ImageNet and two fine-grained datasets: Aircraft and Cars. Our approach consistently improves the performance of Vision Transformers while retaining their properties such as attention to salient regions and higher robustness. Our codes and pre-trained models are available at: https://github.com/hananshafi/vits-for-small-scale-datasets.\",\"PeriodicalId\":72437,\"journal\":{\"name\":\"BMVC : proceedings of the British Machine Vision Conference. British Machine Vision Conference\",\"volume\":\"20 1\",\"pages\":\"731\"},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2022-10-13\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"\",\"citationCount\":\"15\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"BMVC : proceedings of the British Machine Vision Conference. British Machine Vision Conference\",\"FirstCategoryId\":\"1085\",\"ListUrlMain\":\"https://doi.org/10.48550/arXiv.2210.07240\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"\",\"JCRName\":\"\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"BMVC : proceedings of the British Machine Vision Conference. British Machine Vision Conference","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.48550/arXiv.2210.07240","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
How to Train Vision Transformer on Small-scale Datasets?
Vision Transformer (ViT), a radically different architecture than convolutional neural networks offers multiple advantages including design simplicity, robustness and state-of-the-art performance on many vision tasks. However, in contrast to convolutional neural networks, Vision Transformer lacks inherent inductive biases. Therefore, successful training of such models is mainly attributed to pre-training on large-scale datasets such as ImageNet with 1.2M or JFT with 300M images. This hinders the direct adaption of Vision Transformer for small-scale datasets. In this work, we show that self-supervised inductive biases can be learned directly from small-scale datasets and serve as an effective weight initialization scheme for fine-tuning. This allows to train these models without large-scale pre-training, changes to model architecture or loss functions. We present thorough experiments to successfully train monolithic and non-monolithic Vision Transformers on five small datasets including CIFAR10/100, CINIC10, SVHN, Tiny-ImageNet and two fine-grained datasets: Aircraft and Cars. Our approach consistently improves the performance of Vision Transformers while retaining their properties such as attention to salient regions and higher robustness. Our codes and pre-trained models are available at: https://github.com/hananshafi/vits-for-small-scale-datasets.