In recent years, generative adversarial networks (GANs) have been an actively studied topic and shown to successfully produce high-quality realistic images in various domains. The controllable synthesis ability of GAN generators suggests that they maintain informative, disentangled, and explainable image representations, but leveraging and transferring their representations to downstream tasks is largely unexplored. In this paper, we propose to distill knowledge from GAN generators by squeezing and spanning their representations. We squeeze the generator features into representations that are invariant to semantic-preserving transformations through a network before they are distilled into the student network. We span the distilled representation of the synthetic domain to the real domain by also using real training data to remedy the mode collapse of GANs and boost the student network performance in a real domain. Experiments justify the efficacy of our method and reveal its great significance in self-supervised representation learning. Code is available at https://github.com/yangyu12/squeeze-and-span.
translated by 谷歌翻译
Multiview self-supervised representation learning roots in exploring semantic consistency across data of complex intra-class variation. Such variation is not directly accessible and therefore simulated by data augmentations. However, commonly adopted augmentations are handcrafted and limited to simple geometrical and color changes, which are unable to cover the abundant intra-class variation. In this paper, we propose to extract the underlying data variation from datasets and construct a novel augmentation operator, named local manifold augmentation (LMA). LMA is achieved by training an instance-conditioned generator to fit the distribution on the local manifold of data and sampling multiview data using it. LMA shows the ability to create an infinite number of data views, preserve semantics, and simulate complicated variations in object pose, viewpoint, lighting condition, background etc. Experiments show that with LMA integrated, self-supervised learning methods such as MoCov2 and SimSiam gain consistent improvement on prevalent benchmarks including CIFAR10, CIFAR100, STL10, ImageNet100, and ImageNet. Furthermore, LMA leads to representations that obtain more significant invariance to the viewpoint, object pose, and illumination changes and stronger robustness to various real distribution shifts reflected by ImageNet-V2, ImageNet-R, ImageNet Sketch etc.
translated by 谷歌翻译
知识蒸馏在模型压缩方面取得了显着的成就。但是,大多数现有方法需要原始的培训数据,而实践中的实际数据通常是不可用的,因为隐私,安全性和传输限制。为了解决这个问题,我们提出了一种有条件的生成数据无数据知识蒸馏(CGDD)框架,用于培训有效的便携式网络,而无需任何实际数据。在此框架中,除了使用教师模型中提取的知识外,我们将预设标签作为额外的辅助信息介绍以培训发电机。然后,训练有素的发生器可以根据需要产生指定类别的有意义的培训样本。为了促进蒸馏过程,除了使用常规蒸馏损失,我们将预设标签视为地面真理标签,以便学生网络直接由合成训练样本类别监督。此外,我们强制学生网络模仿教师模型的注意图,进一步提高了其性能。为了验证我们方法的优越性,我们设计一个新的评估度量称为相对准确性,可以直接比较不同蒸馏方法的有效性。培训的便携式网络通过提出的数据无数据蒸馏方法获得了99.63%,99.07%和99.84%的CIFAR10,CIFAR100和CALTECH101的相对准确性。实验结果表明了所提出的方法的优越性。
translated by 谷歌翻译
在深度学习研究中,自学学习(SSL)引起了极大的关注,引起了计算机视觉和遥感社区的兴趣。尽管计算机视觉取得了很大的成功,但SSL在地球观测领域的大部分潜力仍然锁定。在本文中,我们对在遥感的背景下为计算机视觉的SSL概念和最新发展提供了介绍,并回顾了SSL中的概念和最新发展。此外,我们在流行的遥感数据集上提供了现代SSL算法的初步基准,从而验证了SSL在遥感中的潜力,并提供了有关数据增强的扩展研究。最后,我们确定了SSL未来研究的有希望的方向的地球观察(SSL4EO),以铺平了两个领域的富有成效的相互作用。
translated by 谷歌翻译
特征回归是将大型神经网络模型蒸馏到较小的功能回归。我们表明,随着网络架构的简单变化,回归可能会优于自我监督模型的知识蒸馏更复杂的最先进方法。令人惊讶的是,即使仅在蒸馏过程中仅使用并且在下游任务中丢弃时,将多层的Perceptron头部添加到CNN骨架上是有益的。因此,更深的非线性投影可以使用在不改变推理架构和时间的情况下准确地模仿老师。此外,我们利用独立的投影头来同时蒸馏多个教师网络。我们还发现,使用与教师和学生网络的输入相同的弱增强图像辅助蒸馏。Imagenet DataSet上的实验证明了各种自我监督蒸馏环境中提出的变化的功效。
translated by 谷歌翻译
知识蒸馏通常涉及如何有效地定义和转移知识从教师到学生。尽管最近的自我监督的对比知识取得了最佳表现,但迫使网络学习此类知识可能会损害对原始班级识别任务的表示。因此,我们采用替代性的自我监督的增强任务来指导网络学习原始识别任务和自我监督的辅助任务的共同分布。它被证明是一种更丰富的知识,可以提高表示能力而不会失去正常的分类能力。此外,以前的方法仅在最终层之间传递概率知识是不完整的。我们建议将几个辅助分类器附加到层次中间特征图中,以生成多样化的自我监督知识,并执行一对一的转移以彻底教授学生网络。我们的方法显着超过了先前的SOTA SSKD,CIFAR-100的平均改善为2.56 \%,并且在广泛使用的网络对上的Imagenet上有0.77 \%的提高。代码可在https://github.com/winycg/hsakd上找到。
translated by 谷歌翻译
尽管自我监督的表示学习(SSL)受到社区的广泛关注,但最近的研究认为,当模型大小降低时,其性能将遭受悬崖的下降。当前的方法主要依赖于对比度学习来训练网络,在这项工作中,我们提出了一种简单而有效的蒸馏对比学习(Disco),以大幅度减轻问题。具体而言,我们发现主流SSL方法获得的最终嵌入包含最富有成果的信息,并建议提炼最终的嵌入,以最大程度地将教师的知识传播到轻量级模型中,通过约束学生的最后嵌入与学生的最后嵌入,以使其与该模型保持一致。老师。此外,在实验中,我们发现存在一种被称为蒸馏瓶颈的现象,并存在以扩大嵌入尺寸以减轻此问题。我们的方法在部署过程中不会向轻型模型引入任何额外的参数。实验结果表明,我们的方法在所有轻型模型上都达到了最先进的作用。特别是,当使用RESNET-101/RESNET-50用作教师教授有效网络-B0时,Imagenet上有效网络B0的线性结果非常接近Resnet-101/Resnet-50,但是有效网络B0的参数数量仅为9.4 \%/16.3 \%Resnet-101/resnet-50。代码可从https:// github获得。 com/yuting-gao/disco-pytorch。
translated by 谷歌翻译
Remarkable progress has been achieved in synthesizing photo-realistic images with generative adversarial networks (GANs). Recently, GANs are utilized as the training sample generator when obtaining or storing real training data is expensive even infeasible. However, traditional GANs generated images are not as informative as the real training samples when being used to train deep neural networks. In this paper, we propose a novel method to synthesize Informative Training samples with GAN (IT-GAN). Specifically, we freeze a pre-trained GAN model and learn the informative latent vectors that correspond to informative training samples. The synthesized images are required to preserve information for training deep neural networks rather than visual reality or fidelity. Experiments verify that the deep neural networks can learn faster and achieve better performance when being trained with our IT-GAN generated images. We also show that our method is a promising solution to dataset condensation problem.
translated by 谷歌翻译
知识蒸馏(KD)是一个有效的框架,旨在将有意义的信息从大型老师转移到较小的学生。通常,KD通常涉及如何定义和转移知识。以前的KD方法通常着重于挖掘各种形式的知识,例如功能地图和精致信息。但是,知识源自主要监督任务,因此是高度特定于任务的。在自我监督的代表学习的最新成功中,我们提出了一项辅助自我实施的增强任务,以指导网络学习更多有意义的功能。因此,我们可以从KD的这项任务中得出软性自我实施的增强分布作为更丰富的黑暗知识。与以前的知识不同,此分布编码从监督和自我监督的特征学习中编码联合知识。除了知识探索之外,我们建议在各个隐藏层上附加几个辅助分支,以充分利用分层特征图。每个辅助分支都被指导学习自学的增强任务,并将这种分布从教师到学生提炼。总体而言,我们称我们的KD方法为等级自我实施的增强知识蒸馏(HSSAKD)。标准图像分类的实验表明,离线和在线HSSAKD都在KD领域达到了最先进的表现。对象检测的进一步转移实验进一步验证了HSSAKD可以指导网络学习更好的功能。该代码可在https://github.com/winycg/hsakd上找到。
translated by 谷歌翻译
近年来有条件的GAN已经成熟,并且能够产生高质量的现实形象。但是,计算资源和培训高质量的GAN所需的培训数据是巨大的,因此对这些模型的转移学习的研究是一个紧急话题。在本文中,我们探讨了从高质量预训练的无条件GAN到有条件的GAN的转移。为此,我们提出了基于HyperNetwork的自适应权重调制。此外,我们介绍了一个自我初始化过程,不需要任何真实数据才能初始化HyperNetwork参数。为了进一步提高知识转移的样本效率,我们建议使用自我监督(对比)损失来改善GaN判别者。在广泛的实验中,我们验证了多个标准基准上的Hypernetworks,自我初始化和对比损失的效率。
translated by 谷歌翻译
基于GAN的生成建模的进展是,社区的推动是为了发现超出图像生成和编辑任务的使用。特别是,最近的几项工作表明,可以重新用诸如零件分割的判别任务重新用来重新用,尤其是当训练数据有限时。但这些改进如何解决自我监督学习的最新进展情况?由此引起这种激励,我们提出了一种基于对比学习的替代方法,并比较它们对标准的几次射击部分分割基准的性能。我们的实验表明,不仅GAN的方法不提供显着的性能优势,它们的多步训练很复杂,几乎是数量级较慢,并且可以引入额外的偏差。这些实验表明,由使用对比学习训练的标准前馈网络捕获的生成模型的感应偏差,例如它们的解开形状和纹理的能力。这些实验表明,目前生成模型中存在的电感偏差,例如它们的解开形状和纹理的能力,通过使用对比学习训练的标准前馈网络充分捕获。
translated by 谷歌翻译
Despite excellent performance in image generation, Generative Adversarial Networks (GANs) are notorious for its requirements of enormous storage and intensive computation. As an awesome ''performance maker'', knowledge distillation is demonstrated to be particularly efficacious in exploring low-priced GANs. In this paper, we investigate the irreplaceability of teacher discriminator and present an inventive discriminator-cooperated distillation, abbreviated as DCD, towards refining better feature maps from the generator. In contrast to conventional pixel-to-pixel match methods in feature map distillation, our DCD utilizes teacher discriminator as a transformation to drive intermediate results of the student generator to be perceptually close to corresponding outputs of the teacher generator. Furthermore, in order to mitigate mode collapse in GAN compression, we construct a collaborative adversarial training paradigm where the teacher discriminator is from scratch established to co-train with student generator in company with our DCD. Our DCD shows superior results compared with existing GAN compression methods. For instance, after reducing over 40x MACs and 80x parameters of CycleGAN, we well decrease FID metric from 61.53 to 48.24 while the current SoTA method merely has 51.92. This work's source code has been made accessible at https://github.com/poopit/DCD-official.
translated by 谷歌翻译
与常规知识蒸馏(KD)不同,自我KD允许网络在没有额外网络的任何指导的情况下向自身学习知识。本文提议从图像混合物(Mixskd)执行自我KD,将这两种技术集成到统一的框架中。 Mixskd相互蒸馏以图形和概率分布在随机的原始图像和它们的混合图像之间以有意义的方式。因此,它通过对混合图像进行监督信号进行建模来指导网络学习跨图像知识。此外,我们通过汇总多阶段功能图来构建一个自学老师网络,以提供软标签以监督骨干分类器,从而进一步提高自我增强的功效。图像分类和转移学习到对象检测和语义分割的实验表明,混合物KD优于其他最先进的自我KD和数据增强方法。该代码可在https://github.com/winycg/self-kd-lib上找到。
translated by 谷歌翻译
生成对抗网络(GAN)具有许多潜在的医学成像应用,包括数据扩展,域适应和模型解释。由于图形处理单元(GPU)的记忆力有限,因此在低分辨率的医学图像上对当前的3D GAN模型进行了训练,因此这些模型要么无法扩展到高分辨率,要么容易出现斑驳的人工制品。在这项工作中,我们提出了一种新颖的端到端GAN体系结构,可以生成高分辨率3D图像。我们通过使用训练和推理之间的不同配置来实现这一目标。在训练过程中,我们采用了层次结构,该结构同时生成图像的低分辨率版本和高分辨率图像的随机选择子量。层次设计具有两个优点:首先,对高分辨率图像训练的记忆需求在子量之间摊销。此外,将高分辨率子体积固定在单个低分辨率图像上可确保子量化之间的解剖一致性。在推断期间,我们的模型可以直接生成完整的高分辨率图像。我们还将具有类似层次结构的编码器纳入模型中,以从图像中提取特征。 3D胸CT和脑MRI的实验表明,我们的方法在图像生成中的表现优于最新技术。我们还证明了所提出的模型在数据增强和临床相关特征提取中的临床应用。
translated by 谷歌翻译
无数据知识蒸馏(DFKD)最近引起了人们的关注,这要归功于其在不使用培训数据的情况下将知识从教师网络转移到学生网络的吸引力。主要思想是使用发电机合成数据以培训学生。随着发电机的更新,合成数据的分布将发生变化。如果发电机和学生接受对手的训练,使学生忘记了先前一步获得的知识,则这种分配转换可能会很大。为了减轻这个问题,我们提出了一种简单而有效的方法,称为动量对抗蒸馏(MAD),该方法维持了发电机的指数移动平均值(EMA)副本,并使用发电机和EMA生成器的合成样品来培训学生。由于EMA发电机可以被视为发电机旧版本的合奏,并且与发电机相比,更新的更改通常会发生较小的变化,因此对其合成样本进行培训可以帮助学生回顾过去的知识,并防止学生适应太快的速度发电机的新更新。我们在六个基准数据集上进行的实验,包括ImageNet和Place365,表明MAD的性能优于竞争方法来处理大型分配转移问题。我们的方法还与现有的DFKD方法相比,甚至在某些情况下达到了最新的方法。
translated by 谷歌翻译
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.
translated by 谷歌翻译
Often we wish to transfer representational knowledge from one neural network to another. Examples include distilling a large network into a smaller one, transferring knowledge from one sensory modality to a second, or ensembling a collection of models into a single estimator. Knowledge distillation, the standard approach to these problems, minimizes the KL divergence between the probabilistic outputs of a teacher and student network. We demonstrate that this objective ignores important structural knowledge of the teacher network. This motivates an alternative objective by which we train a student to capture significantly more information in the teacher's representation of the data. We formulate this objective as contrastive learning. Experiments demonstrate that our resulting new objective outperforms knowledge distillation and other cutting-edge distillers on a variety of knowledge transfer tasks, including single model compression, ensemble distillation, and cross-modal transfer. Our method sets a new state-of-the-art in many transfer tasks, and sometimes even outperforms the teacher network when combined with knowledge distillation.
translated by 谷歌翻译
The success of state-of-the-art deep neural networks heavily relies on the presence of large-scale labelled datasets, which are extremely expensive and time-consuming to annotate. This paper focuses on tackling semi-supervised part segmentation tasks by generating high-quality images with a pre-trained GAN and labelling the generated images with an automatic annotator. In particular, we formulate the annotator learning as a learning-to-learn problem. Given a pre-trained GAN, the annotator learns to label object parts in a set of randomly generated images such that a part segmentation model trained on these synthetic images with their predicted labels obtains low segmentation error on a small validation set of manually labelled images. We further reduce this nested-loop optimization problem to a simple gradient matching problem and efficiently solve it with an iterative algorithm. We show that our method can learn annotators from a broad range of labelled images including real images, generated images, and even analytically rendered images. Our method is evaluated with semi-supervised part segmentation tasks and significantly outperforms other semi-supervised competitors when the amount of labelled examples is extremely limited.
translated by 谷歌翻译
具有高计算成本的生成对抗网络(GANS),例如Biggan和Stylegan2,实现了显着的结果,在随机噪声中合成高分辨率和多样化的图像。降低GAN的计算成本,同时保持发电照片逼真的图像是一种紧急和具有挑战性的领域,用于其在计算资源限制设备上的广泛应用。在这项工作中,我们提出了一种新颖又简单的{\ bf d} isCriminator {\ bf g} uided {\ bf l}用于压缩vanilla {\ bf gaN}的折射方法,称为{\ bf dgl-gan}。受到教师歧视者可能包含一些有意义信息的现象的动机,我们通过对抗函数从教师歧视者转移知识。我们展示DGL-GAN自体虚拟性有效,从教师歧视者学习可以促进学生会的表现,通过广泛的实验结果验证。此外,我们提出了一个两级培训DGL-GAN的培训策略,当我们申请DGL-GAN来压缩两种最具代表性大规模的Vanilla Gans时,可以大大稳定其培训过程并实现卓越的性能。 。实验表明,DGL-GAN实现了最先进的(SOTA)在STYLAG2(FFHQ上的FID 2.92上有近1/3 $参数的FFH3)和Biggan(93.29和FID 9.92,在想象中有近1美元/ Biggan的4 $参数)并优于几种现有的香草GAN压缩技术。此外,DGL-GAN也有效地提高了原始未压缩的GAN的性能,原始未压缩的风格2升高的DGL-GAN促进了FFHQ的FID 2.65,这实现了新的最先进的性能。代码和模型可用于\ url {https://github.com/yuesongtian/dgl-gan}。
translated by 谷歌翻译
基础模型不是模型生产管道的最后一章。以少数数据以少数数据传输到数千个下游任务正在成为基础模型的应用的趋势。在本文中,我们提出了一个通用转移框架:一个传输所有(OTA),将任何视觉基础模型(VFM)转移到具有少数下游数据的下游任务。我们首先通过图像重新表示微调(IRF)将VFM传输到特定于任务特定模型,然后将知识从特定于任务的模型蒸馏到部署的模型,其中包含由下游图像引导的生成(DIGG)产生的数据。OTA在传输时没有对上游数据,VFM和下游任务的依赖性。它还为VFM研究人员提供了一种方法,以释放其上游信息,以便更好地转移,但由于隐私要求而没有泄漏数据。大规模实验在少数数据设置中验证我们方法的有效性和优越性。我们的代码将被释放。
translated by 谷歌翻译