Transfer learning from natural image datasets, particularly I N , using standard large models and corresponding pretrained weights has become a de-facto method for deep learning applications to medical imaging. However, there are fundamental di erences in data sizes, features and task speci cations between natural image classi cation and the target medical tasks, and there is little understanding of the e ects of transfer. In this paper, we explore properties of transfer learning for medical imaging. A performance evaluation on two large scale medical imaging tasks shows that surprisingly, transfer o ers little bene t to performance, and simple, lightweight models can perform comparably to I N architectures. Investigating the learned representations and features, we nd that some of the di erences from transfer learning are due to the over-parametrization of standard models rather than sophisticated feature reuse. We isolate where useful feature reuse occurs, and outline the implications for more e cient model exploration. We also explore feature independent bene ts of transfer arising from weight scalings.

转移学习是一种标准技术，可以将知识从一个领域转移到另一个领域。对于医学成像中的应用，尽管域之间的任务和图像特征差异，但从Imagenet转移已成为事实上的方法。但是，尚不清楚哪些因素决定了哪些因素以及在何种程度上转移学习到医疗领域是有用的。最近，人们对源域重复使用的特征的长期假设最近受到质疑。通过在几个医学图像基准数据集上进行的一系列实验，我们探讨了传输学习，数据大小，模型的容量和电感偏置以及源域和目标域之间的距离之间的关系。我们的发现表明，在大多数情况下，转移学习是有益的，我们表征了重要的角色重复使用在其成功方面。

Transfer learning is a cornerstone of computer vision, yet little work has been done to evaluate the relationship between architecture and transfer. An implicit hypothesis in modern computer vision research is that models that perform better on ImageNet necessarily perform better on other vision tasks. However, this hypothesis has never been systematically tested. Here, we compare the performance of 16 classification networks on 12 image classification datasets. We find that, when networks are used as fixed feature extractors or fine-tuned, there is a strong correlation between ImageNet accuracy and transfer accuracy (r = 0.99 and 0.96, respectively). In the former setting, we find that this relationship is very sensitive to the way in which networks are trained on ImageNet; many common forms of regularization slightly improve ImageNet accuracy but yield penultimate layer features that are much worse for transfer learning. Additionally, we find that, on two small fine-grained image classification datasets, pretraining on ImageNet provides minimal benefits, indicating the learned features from Ima-geNet do not transfer well to fine-grained tasks. Together, our results show that ImageNet architectures generalize well across datasets, but ImageNet features are less general than previously suggested.

以前的工作提出了许多新的损失函数和常规程序，可提高图像分类任务的测试准确性。但是，目前尚不清楚这些损失函数是否了解下游任务的更好表示。本文研究了培训目标的选择如何影响卷积神经网络隐藏表示的可转移性，训练在想象中。我们展示了许多目标在Vanilla Softmax交叉熵上导致想象的精度有统计学意义的改进，但由此产生的固定特征提取器转移到下游任务基本较差，并且当网络完全微调时，损失的选择几乎没有效果新任务。使用居中内核对齐来测量网络隐藏表示之间的相似性，我们发现损失函数之间的差异仅在网络的最后几层中都很明显。我们深入了解倒数第二层的陈述，发现不同的目标和近奇计的组合导致大幅不同的类别分离。具有较高类别分离的表示可以在原始任务上获得更高的准确性，但它们的功能对于下游任务不太有用。我们的结果表明，用于原始任务的学习不变功能与传输任务相关的功能之间存在权衡。

Transfer of pre-trained representations improves sample efficiency and simplifies hyperparameter tuning when training deep neural networks for vision. We revisit the paradigm of pre-training on large supervised datasets and fine-tuning the model on a target task. We scale up pre-training, and propose a simple recipe that we call Big Transfer (BiT). By combining a few carefully selected components, and transferring using a simple heuristic, we achieve strong performance on over 20 datasets. BiT performs well across a surprisingly wide range of data regimes -from 1 example per class to 1 M total examples. BiT achieves 87.5% top-1 accuracy on ILSVRC-2012, 99.4% on CIFAR-10, and 76.3% on the 19 task Visual Task Adaptation Benchmark (VTAB). On small datasets, BiT attains 76.8% on ILSVRC-2012 with 10 examples per class, and 97.0% on CIFAR-10 with 10 examples per class. We conduct detailed analysis of the main components that lead to high transfer performance.

Convolutional neural networks (CNNs) have so far been the de-facto model for visual data. Recent work has shown that (Vision) Transformer models (ViT) can achieve comparable or even superior performance on image classification tasks. This raises a central question: how are Vision Transformers solving these tasks? Are they acting like convolutional networks, or learning entirely different visual representations? Analyzing the internal representation structure of ViTs and CNNs on image classification benchmarks, we find striking differences between the two architectures, such as ViT having more uniform representations across all layers. We explore how these differences arise, finding crucial roles played by self-attention, which enables early aggregation of global information, and ViT residual connections, which strongly propagate features from lower to higher layers. We study the ramifications for spatial localization, demonstrating ViTs successfully preserve input spatial information, with noticeable effects from different classification methods. Finally, we study the effect of (pretraining) dataset scale on intermediate features and transfer learning, and conclude with a discussion on connections to new architectures such as the MLP-Mixer. This breakthrough highlights a fundamental question: how are Vision Transformers solving these image based tasks? Do they act like convolutions, learning the same inductive biases from scratch? Or are they developing novel task representations? What is the role of scale in learning these representations? And are there ramifications for downstream tasks? In this paper, we study these questions, uncovering key representational differences between ViTs and CNNs, the ways in which these difference arise, and effects on classification and transfer learning. Specifically, our contributions are:35th Conference on Neural Information Processing Systems (NeurIPS 2021).

转移学习可以在源任务上重新使用知识来帮助学习目标任务。一种简单的转移学习形式在当前的最先进的计算机视觉模型中是常见的，即预先训练ILSVRC数据集上的图像分类模型，然后在任何目标任务上进行微调。然而，先前对转移学习的系统研究已经有限，并且预计工作的情况并不完全明白。在本文中，我们对跨越不同的图像域进行了广泛的转移学习实验探索（消费者照片，自主驾驶，空中图像，水下，室内场景，合成，特写镜头）和任务类型（语义分割，物体检测，深度估计，关键点检测）。重要的是，这些都是与现代计算机视觉应用相关的复杂的结构化的输出任务类型。总共执行超过2000年的转移学习实验，包括许多来源和目标来自不同的图像域，任务类型或两者。我们系统地分析了这些实验，了解图像域，任务类型和数据集大小对传输学习性能的影响。我们的研究导致了几个见解和具体建议：（1）对于大多数任务，存在一个显着优于ILSVRC'12预培训的来源; （2）图像领域是实现阳性转移的最重要因素; （3）源数据集应该\ \ emph {include}目标数据集的图像域以获得最佳结果; （4）与此同时，当源任务的图像域比目标的图像域时，我们只观察小的负面影响; （5）跨任务类型的转移可能是有益的，但其成功严重依赖于源和目标任务类型。

Vision transformer (ViT) models exhibit substandard optimizability. In particular, they are sensitive to the choice of optimizer (AdamW vs. SGD), optimizer hyperparameters, and training schedule length. In comparison, modern convolutional neural networks are easier to optimize. Why is this the case? In this work, we conjecture that the issue lies with the patchify stem of ViT models, which is implemented by a stride-p p×p convolution (p = 16 by default) applied to the input image. This large-kernel plus large-stride convolution runs counter to typical design choices of convolutional layers in neural networks. To test whether this atypical design choice causes an issue, we analyze the optimization behavior of ViT models with their original patchify stem versus a simple counterpart where we replace the ViT stem by a small number of stacked stride-two 3×3 convolutions. While the vast majority of computation in the two ViT designs is identical, we find that this small change in early visual processing results in markedly different training behavior in terms of the sensitivity to optimization settings as well as the final model accuracy. Using a convolutional stem in ViT dramatically increases optimization stability and also improves peak performance (by ∼1-2% top-1 accuracy on ImageNet-1k), while maintaining flops and runtime. The improvement can be observed across the wide spectrum of model complexities (from 1G to 36G flops) and dataset scales (from ImageNet-1k to ImageNet-21k). These findings lead us to recommend using a standard, lightweight convolutional stem for ViT models in this regime as a more robust architectural choice compared to the original ViT model design.

最近，已经观察到，转移学习解决方案可能是我们解决许多少量学习基准的全部 - 因此提出了有关何时以及如何部署元学习算法的重要问题。在本文中，我们试图通过1.提出一个新颖的指标（多样性系数）来阐明这些问题，以测量几次学习基准和2.的任务多样性。 ）并在公平条件下进行学习（相同的体系结构，相同的优化器和所有经过培训的模型）。使用多样性系数，我们表明流行的迷你胶原和Cifar-fs几乎没有学习基准的多样性低。这种新颖的洞察力将转移学习解决方案比在公平比较的低多样性方面的元学习解决方案更好。具体而言，我们从经验上发现，低多样性系数与转移学习和MAML学习解决方案之间的高相似性在元测试时间和分类层相似性方面（使用基于特征的距离指标，例如SVCCA，PWCCA，CKA和OPD） ）。为了进一步支持我们的主张，我们发现这种元测试的准确性仍然存在，即使模型大小变化也是如此。因此，我们得出的结论是，在低多样性制度中，MAML和转移学习在公平比较时具有等效的元检验性能。我们也希望我们的工作激发了对元学习基准测试基准的更周到的结构和定量评估。

Transfer learning (TL) from pretrained deep models is a standard practice in modern medical image classification (MIC). However, what levels of features to be reused are problem-dependent, and uniformly finetuning all layers of pretrained models may be suboptimal. This insight has partly motivated the recent \emph{differential} TL strategies, such as TransFusion (TF) and layer-wise finetuning (LWFT), which treat the layers in the pretrained models differentially. In this paper, we add one more strategy into this family, called \emph{TruncatedTL}, which reuses and finetunes appropriate bottom layers and directly discards the remaining layers. This yields not only superior MIC performance but also compact models for efficient inference, compared to other differential TL methods. We validate the performance and model efficiency of TruncatedTL on three MIC tasks covering both 2D and 3D images. For example, on the BIMCV COVID-19 classification dataset, we obtain improved performance with around $1/4$ model size and $2/3$ inference time compared to the standard full TL model. Code is available at https://github.com/sun-umn/Transfer-Learning-in-Medical-Imaging.

Driven by improved architectures and better representation learning frameworks, the field of visual recognition has enjoyed rapid modernization and performance boost in the early 2020s. For example, modern ConvNets, represented by ConvNeXt, have demonstrated strong performance in various scenarios. While these models were originally designed for supervised learning with ImageNet labels, they can also potentially benefit from self-supervised learning techniques such as masked autoencoders (MAE). However, we found that simply combining these two approaches leads to subpar performance. In this paper, we propose a fully convolutional masked autoencoder framework and a new Global Response Normalization (GRN) layer that can be added to the ConvNeXt architecture to enhance inter-channel feature competition. This co-design of self-supervised learning techniques and architectural improvement results in a new model family called ConvNeXt V2, which significantly improves the performance of pure ConvNets on various recognition benchmarks, including ImageNet classification, COCO detection, and ADE20K segmentation. We also provide pre-trained ConvNeXt V2 models of various sizes, ranging from an efficient 3.7M-parameter Atto model with 76.7% top-1 accuracy on ImageNet, to a 650M Huge model that achieves a state-of-the-art 88.9% accuracy using only public training data.

Traditionally, deep learning methods for breast cancer classification perform a single-view analysis. However, radiologists simultaneously analyze all four views that compose a mammography exam, owing to the correlations contained in mammography views, which present crucial information for identifying tumors. In light of this, some studies have started to propose multi-view methods. Nevertheless, in such existing architectures, mammogram views are processed as independent images by separate convolutional branches, thus losing correlations among them. To overcome such limitations, in this paper we propose a novel approach for multi-view breast cancer classification based on parameterized hypercomplex neural networks. Thanks to hypercomplex algebra properties, our networks are able to model, and thus leverage, existing correlations between the different views that comprise a mammogram, thus mimicking the reading process performed by clinicians. The proposed methods are able to handle the information of a patient altogether without breaking the multi-view nature of the exam. We define architectures designed to process two-view exams, namely PHResNets, and four-view exams, i.e., PHYSEnet and PHYBOnet. Through an extensive experimental evaluation conducted with publicly available datasets, we demonstrate that our proposed models clearly outperform real-valued counterparts and also state-of-the-art methods, proving that breast cancer classification benefits from the proposed multi-view architectures. We also assess the method's robustness beyond mammogram analysis by considering different benchmarks, as well as a finer-scaled task such as segmentation. Full code and pretrained models for complete reproducibility of our experiments are freely available at: https://github.com/ispamm/PHBreast.

Neural networks require careful weight initialization to prevent signals from exploding or vanishing. Existing initialization schemes solve this problem in specific cases by assuming that the network has a certain activation function or topology. It is difficult to derive such weight initialization strategies, and modern architectures therefore often use these same initialization schemes even though their assumptions do not hold. This paper introduces AutoInit, a weight initialization algorithm that automatically adapts to different neural network architectures. By analytically tracking the mean and variance of signals as they propagate through the network, AutoInit appropriately scales the weights at each layer to avoid exploding or vanishing signals. Experiments demonstrate that AutoInit improves performance of convolutional, residual, and transformer networks across a range of activation function, dropout, weight decay, learning rate, and normalizer settings, and does so more reliably than data-dependent initialization methods. This flexibility allows AutoInit to initialize models for everything from small tabular tasks to large datasets such as ImageNet. Such generality turns out particularly useful in neural architecture search and in activation function discovery. In these settings, AutoInit initializes each candidate appropriately, making performance evaluations more accurate. AutoInit thus serves as an automatic configuration tool that makes design of new neural network architectures more robust. The AutoInit package provides a wrapper around TensorFlow models and is available at https://github.com/cognizant-ai-labs/autoinit.

通过卫星摄像机获取关于地球表面的大面积的信息使我们能够看到远远超过我们在地面上看到的更多。这有助于我们在检测和监测土地使用模式，大气条件，森林覆盖和许多非上市方面的地区的物理特征。所获得的图像不仅跟踪连续的自然现象，而且对解决严重森林砍伐的全球挑战也至关重要。其中亚马逊盆地每年占最大份额。适当的数据分析将有助于利用可持续健康的氛围来限制对生态系统和生物多样性的不利影响。本报告旨在通过不同的机器学习和优越的深度学习模型用大气和各种陆地覆盖或土地使用亚马逊雨林的卫星图像芯片。评估是基于F2度量完成的，而用于损耗函数，我们都有S形跨熵以及Softmax交叉熵。在使用预先训练的ImageNet架构中仅提取功能之后，图像被间接馈送到机器学习分类器。鉴于深度学习模型，通过传输学习使用微调Imagenet预训练模型的集合。到目前为止，我们的最佳分数与F2度量为0.927。

Increasing model, data and compute budget scale in the pre-training has been shown to strongly improve model generalization and transfer learning in vast line of work done in language modeling and natural image recognition. However, most studies on the positive effect of larger scale were done in scope of in-domain setting, with source and target data being in close proximity. To study effect of larger scale for both in-domain and out-of-domain setting when performing full and few-shot transfer, we combine here for the first time large, openly available medical X-Ray chest imaging datasets to reach a scale for medical imaging domain comparable to ImageNet-1k, routinely used for pre-training in natural image domain. We then conduct supervised pre-training, while varying network size and source data scale and domain, being either large natural (ImageNet-1k/21k) or large medical chest X-Ray datasets, and transfer pre-trained models to different natural or medical targets. We observe strong improvement due to larger pre-training scale for intra-domain natural-natural and medical-medical transfer. For inter-domain natural-medical transfer, we find improvements due to larger pre-training scale on larger X-Ray targets in full shot regime, while for smaller targets and for few-shot regime the improvement is not visible. Remarkably, large networks pre-trained on very large natural ImageNet-21k are as good or better than networks pre-trained on largest available medical X-Ray data when performing transfer to large X-Ray targets. We conclude that substantially increasing model and generic, medical domain-agnostic natural image source data scale in the pre-training can enable high quality out-of-domain transfer to medical domain specific targets, removing dependency on large medical domain-specific source data often not available in the practice.

神经架构的创新促进了语言建模和计算机视觉中的重大突破。不幸的是，如果网络参数未正确初始化，新颖的架构通常会导致挑战超参数选择和培训不稳定。已经提出了许多架构特定的初始化方案，但这些方案并不总是可移植到新体系结构。本文介绍了毕业，一种用于初始化神经网络的自动化和架构不可知论由方法。毕业基础是一个简单的启发式;调整每个网络层的规范，使得具有规定的超参数的SGD或ADAM的单个步骤导致可能的损耗值最小。通过在每个参数块前面引入标量乘数变量，然后使用简单的数字方案优化这些变量来完成此调整。 GradInit加速了许多卷积架构的收敛性和测试性能，无论是否有跳过连接，甚至没有归一化层。它还提高了机器翻译的原始变压器架构的稳定性，使得在广泛的学习速率和动量系数下使用ADAM或SGD来训练它而无需学习速率预热。代码可在https://github.com/zhuchen03/gradinit上获得。

使用DataSet的真实标签培训而不是随机标签导致更快的优化和更好的泛化。这种差异归因于自然数据集中的输入和标签之间的对齐概念。我们发现，随机或真正标签上的具有不同架构和优化器的培训神经网络在隐藏的表示和训练标签之间强制执行相同的关系，阐明为什么神经网络表示为转移如此成功。我们首先突出显示为什么对齐的特征在经典的合成转移问题中促进转移和展示，即对齐是对相似和不同意任务的正负传输的确定因素。然后我们调查各种神经网络架构，并发现（a）在各种不同的架构和优化器中出现的对齐，并且从深度（b）对准产生的更多对准对于更接近输出的层和（c）现有的性能深度CNN表现出高级别的对准。

Neural network pruning-the task of reducing the size of a network by removing parameters-has been the subject of a great deal of work in recent years. We provide a meta-analysis of the literature, including an overview of approaches to pruning and consistent findings in the literature. After aggregating results across 81 papers and pruning hundreds of models in controlled conditions, our clearest finding is that the community suffers from a lack of standardized benchmarks and metrics. This deficiency is substantial enough that it is hard to compare pruning techniques to one another or determine how much progress the field has made over the past three decades. To address this situation, we identify issues with current practices, suggest concrete remedies, and introduce ShrinkBench, an open-source framework to facilitate standardized evaluations of pruning methods. We use ShrinkBench to compare various pruning techniques and show that its comprehensive evaluation can prevent common pitfalls when comparing pruning methods.

We introduce Bootstrap Your Own Latent (BYOL), a new approach to self-supervised image representation learning. BYOL relies on two neural networks, referred to as online and target networks, that interact and learn from each other. From an augmented view of an image, we train the online network to predict the target network representation of the same image under a different augmented view. At the same time, we update the target network with a slow-moving average of the online network. While state-of-the art methods rely on negative pairs, BYOL achieves a new state of the art without them. BYOL reaches 74.3% top-1 classification accuracy on ImageNet using a linear evaluation with a ResNet-50 architecture and 79.6% with a larger ResNet. We show that BYOL performs on par or better than the current state of the art on both transfer and semi-supervised benchmarks. Our implementation and pretrained models are given on GitHub. 3 * Equal contribution; the order of first authors was randomly selected.

This paper presents SimCLR: a simple framework for contrastive learning of visual representations. We simplify recently proposed contrastive selfsupervised learning algorithms without requiring specialized architectures or a memory bank. In order to understand what enables the contrastive prediction tasks to learn useful representations, we systematically study the major components of our framework. We show that (1) composition of data augmentations plays a critical role in defining effective predictive tasks, (2) introducing a learnable nonlinear transformation between the representation and the contrastive loss substantially improves the quality of the learned representations, and (3) contrastive learning benefits from larger batch sizes and more training steps compared to supervised learning. By combining these findings, we are able to considerably outperform previous methods for self-supervised and semi-supervised learning on ImageNet. A linear classifier trained on self-supervised representations learned by Sim-CLR achieves 76.5% top-1 accuracy, which is a 7% relative improvement over previous state-ofthe-art, matching the performance of a supervised ResNet-50. When fine-tuned on only 1% of the labels, we achieve 85.8% top-5 accuracy, outperforming AlexNet with 100× fewer labels. 1