Learning with noisy-labels has become an important research topic in computer vision where state-of-the-art (SOTA) methods explore: 1) prediction disagreement with co-teaching strategy that updates two models when they disagree on the prediction of training samples; and 2) sample selection to divide the training set into clean and noisy sets based on small training loss. However, the quick convergence of co-teaching models to select the same clean subsets combined with relatively fast overfitting of noisy labels may induce the wrong selection of noisy label samples as clean, leading to an inevitable confirmation bias that damages accuracy. In this paper, we introduce our noisy-label learning approach, called Asymmetric Co-teaching (AsyCo), which introduces novel prediction disagreement that produces more consistent divergent results of the co-teaching models, and a new sample selection approach that does not require small-loss assumption to enable a better robustness to confirmation bias than previous methods. More specifically, the new prediction disagreement is achieved with the use of different training strategies, where one model is trained with multi-class learning and the other with multi-label learning. Also, the new sample selection is based on multi-view consensus, which uses the label views from training labels and model predictions to divide the training set into clean and noisy for training the multi-class model and to re-label the training samples with multiple top-ranked labels for training the multi-label model. Extensive experiments on synthetic and real-world noisy-label datasets show that AsyCo improves over current SOTA methods.
translated by 谷歌翻译
The importance of learning rate (LR) schedules on network pruning has been observed in a few recent works. As an example, Frankle and Carbin (2019) highlighted that winning tickets (i.e., accuracy preserving subnetworks) can not be found without applying a LR warmup schedule and Renda, Frankle and Carbin (2020) demonstrated that rewinding the LR to its initial state at the end of each pruning cycle improves performance. In this paper, we go one step further by first providing a theoretical justification for the surprising effect of LR schedules. Next, we propose a LR schedule for network pruning called SILO, which stands for S-shaped Improved Learning rate Optimization. The advantages of SILO over existing state-of-the-art (SOTA) LR schedules are two-fold: (i) SILO has a strong theoretical motivation and dynamically adjusts the LR during pruning to improve generalization. Specifically, SILO increases the LR upper bound (max_lr) in an S-shape. This leads to an improvement of 2% - 4% in extensive experiments with various types of networks (e.g., Vision Transformers, ResNet) on popular datasets such as ImageNet, CIFAR-10/100. (ii) In addition to the strong theoretical motivation, SILO is empirically optimal in the sense of matching an Oracle, which exhaustively searches for the optimal value of max_lr via grid search. We find that SILO is able to precisely adjust the value of max_lr to be within the Oracle optimized interval, resulting in performance competitive with the Oracle with significantly lower complexity.
translated by 谷歌翻译
The proliferation of smartphones has accelerated mobility studies by largely increasing the type and volume of mobility data available. One such source of mobility data is from GPS technology, which is becoming increasingly common and helps the research community understand mobility patterns of people. However, there lacks a standardized framework for studying the different mobility patterns created by the non-Work, non-Home locations of Working and Nonworking users on Workdays and Offdays using machine learning methods. We propose a new mobility metric, Daily Characteristic Distance, and use it to generate features for each user together with Origin-Destination matrix features. We then use those features with an unsupervised machine learning method, $k$-means clustering, and obtain three clusters of users for each type of day (Workday and Offday). Finally, we propose two new metrics for the analysis of the clustering results, namely User Commonality and Average Frequency. By using the proposed metrics, interesting user behaviors can be discerned and it helps us to better understand the mobility patterns of the users.
translated by 谷歌翻译
最先进的(SOTA)深度学习乳房X线照片分类器接受了弱标记的图像训练,通常依赖于产生有限解释性预测的全球模型,这是他们成功地转化为临床实践的关键障碍。另一方面,基于原型的模型通过将预测与训练图像原型相关联,改善了可解释性,但是它们的准确性不如全球模型,其原型往往具有差的多样性。我们通过BraixProtopnet ++的建议解决了这两个问题,该问题通过将基于原型的模型结合起来,为全局模型增添了解释性。 BraixProtopnet ++在训练基于原型的模型以提高合奏的分类精度时,会提炼全局模型的知识。此外,我们提出了一种方法来通过保证所有原型都与不同的训练图像相关联,以增加原型多样性。对弱标记的私人和公共数据集进行的实验表明,BraixProtopnet ++的分类精度比基于SOTA Global和基于原型的模型具有更高的分类精度。使用病变定位来评估模型可解释性,我们显示BraixProtopnet ++比其他基于原型的模型和全球模型的事后解释更有效。最后,我们表明,BraixProtopnet ++学到的原型的多样性优于基于SOTA原型的方法。
translated by 谷歌翻译
在分析筛查乳房X线照片时,放射科医生可以自然处理每个乳房的两个同侧视图,即颅底审计(CC)和中外侧 - 粘合剂(MLO)视图。这些多个相关图像提供了互补的诊断信息,并可以提高放射科医生的分类准确性。不幸的是,大多数现有的深度学习系统,受过全球标记的图像培训,缺乏从这些多种观点中共同分析和整合全球和本地信息的能力。通过忽略筛选发作的多个图像中存在的潜在有价值的信息,人们限制了这些系统的潜在准确性。在这里,我们提出了一种新的多视图全球分析方法,该方法基于全球一致性学习和对乳房X线照片中同侧观点的局部同时学习,模仿放射科医生的阅读程序。广泛的实验表明,在大规模的私人数据集和两个公开可用的数据集上,我们的模型在分类准确性和概括方面优于竞争方法,在该数据集和两个公开可用的数据集上,模型仅受到全球标签的培训和测试。
translated by 谷歌翻译
本文介绍了Omnivl,这是一种新的基础模型,旨在使用一种通用体系结构来支持图像语言和视频语言任务。它为图像和视频输入采用了统一的基于变压器的视觉编码器,因此可以执行联合图像语言和视频语言预处理。我们首次证明了这样的范式受益于图像和视频任务,而不是传统的单向传输(例如,使用图像语言来帮助视频语言)。为此,我们提出了对图像语言和视频语言的脱钩关节预处理,以有效地将视觉模型分解为空间和时间维度,并在图像和视频任务上获得性能提升。此外,我们引入了一种新颖的统一视觉对比度(UNIVLC)损失,以利用图像文本,视频文本,图像标签(例如,图像分类),视频标签(例如,视频动作识别)在一起受到监督和吵闹的监督预处理数据都尽可能多地利用。无需额外的任务适配器,Omnivl可以同时支持仅视觉任务(例如,图像分类,视频操作识别),跨模式对齐任务(例如,图像/视频 - 文本检索)和多模式理解和生成任务(例如,图像/视频问答,字幕)。我们在各种下游任务上评估Omnivl,并以相似的模型大小和数据量表获得最新的或竞争结果。
translated by 谷歌翻译
零拍学习(ZSL)旨在通过利用所见类和看不见的类之间共享的语义描述来识别看不见的类。当前的方法表明,通过将语义嵌入将视觉空间投射到视觉空间中是类原型,从而有效地学习视觉语义对齐是有效的。但是,这样的投影函数仅与可见的类有关。当应用于看不见的类时,原型通常由于域移位而次优。在本文中,我们建议通过称为LPL的占位符学习原型,以消除看到和看不见的阶级之间的域转移。具体来说,我们将看到的课程结合在一起,以使新课程成为视觉和语义空间中看不见的班级的占位符。占位持有人放置在看到的班级之间,鼓励人们高度分散所见类的原型。插入良好的看不见的空间也可以保留更多的空间。从经验上讲,分离良好的原型有助于抵消由域转移引起的视觉声音错位。此外,我们利用一种新颖的面向语义的微调来保证占位符的语义可靠性。在五个基准数据集上进行的广泛实验证明了LPL在最新方法上的显着性能提高。代码可在https://github.com/zaiquanyang/lpl上找到。
translated by 谷歌翻译
如今,无线通信正在迅速重塑整个行业。特别是,移动边缘计算(MEC)是一种用于工业互联网(IIOT)的促成技术,它使强大的计算/存储基础架构更靠近移动终端,从而大大降低了响应延迟。为了获得在网络边缘积极缓存的好处,对最终设备之间的受欢迎程度的精确知识至关重要。但是,在许多IIOT场景中,内容流行的内容流行以及数据私人关系的复杂性质对其获取构成了艰巨的挑战。在本文中,我们建议针对MEC启用的IIOT提供无监督和保护隐私的普及预测框架。引入了本地和全球流行的概念,并将每个用户的随时间变化为无模型的马尔可夫链。在此基础上,提出了一种新颖的无监督的复发性联合学习(URFL)算法,以预测分布式的流行,同时实现隐私保护和无监督的培训。仿真表明,提出的框架可以根据降低的根平方误差提高预测准确性,高达$ 60.5 \%-68.7 \%$。此外,避免了手动标签和违反用户数据隐私的行为。
translated by 谷歌翻译
快捷方式学习对深度学习模型很常见,但导致了退化的特征表示形式,因此危害了该模型的可推广性和解释性。但是,在广泛使用的视觉变压器框架中的快捷方式学习在很大程度上是未知的。同时,引入特定领域的知识是纠正捷径的主要方法,捷径为背景相关因素。例如,在医学成像领域中,放射科医生的眼睛凝视数据是一种有效的人类视觉先验知识,具有指导深度学习模型的巨大潜力,可以专注于有意义的前景区域。但是,获得眼睛凝视数据是时必的,劳动密集型的,有时甚至是不切实际的。在这项工作中,我们提出了一种新颖而有效的显着性视觉变压器(SGT)模型,以在没有眼神数据的情况下在VIT中纠正快捷方式学习。具体而言,采用计算视觉显着性模型来预测输入图像样本的显着性图。然后,显着图用于散布最有用的图像贴片。在拟议的中士中,图像贴片之间的自我注意力仅集中于蒸馏的信息。考虑到这种蒸馏操作可能会导致全局信息丢失,我们在最后一个编码器层中进一步介绍了一个残留的连接,该连接捕获了所有图像贴片中的自我注意力。四个独立公共数据集的实验结果表明,我们的SGT框架可以有效地学习和利用人类的先验知识,而无需眼睛凝视数据,并且比基线更好。同时,它成功地纠正了有害的快捷方式学习并显着提高了VIT模型的解释性,证明了传递人类先验知识在纠正快捷方式学习方面传递人类先验知识的承诺
translated by 谷歌翻译
差异化私有(DP)数据发布是一种有前途的技术,可以在不损害数据主体的隐私而传播数据。但是,大多数先前的工作都集中在单一方拥有所有数据的方案上。在本文中,我们专注于多方设置,其中不同的利益相关者拥有属于同一数据主体的属性集合。在线性回归的上下文中,允许各方在完全数据上训练模型,而无需推断个人的私人属性或身份,我们首先直接应用高斯机制并表明其具有小的特征值问题。我们进一步提出了我们的新方法,并证明其渐近地收敛到随着数据集大小增加的最佳(非私有)解决方案。我们通过对人工和现实世界数据集的实验来证实理论结果。
translated by 谷歌翻译