Two-way imputation: A Bayesian method for estimating missing scores in tests and questionnaires, and an accurate approximation

Previous research has shown that method two-way with error for multiple imputation in test and questionnaire data produces small bias in statistical analyses. This method is based on a two-way ANOVA model of persons by items but it is improper from a Bayesian point of view. Proper two-way imputations are generated using data augmentation. Simulation results show that the resulting method two-way with data augmentation produces unbiased results in Cronbach's alpha, the mean of squares in ANOVA, the item means, and small bias in the mean test score and the factor loadings from principal components analysis. The data with imputed scores result in statistics having a slightly larger standard deviation than the original complete data. Method two-way with error produces results that are only slightly more biased, especially for low percentages of missingness. Thus, it may serve as an accurate approximation to the more involved method two-way with data augmentation.     

机器视觉应用中的图像数据增广综述

深度学习是目前机器视觉的前沿解决方案,而海量高质量的训练数据集是深度学习解决机器视觉问题的基本保障.收集和准确标注图像数据集是一个极其费时且代价昂贵的过程.随着机器视觉的广泛应用,这个问题将会越来越突出.图像增广技术是一种有效解决深度学习在少量或者低质量训练数据中进行训练的一种技术手段,该技术不断地伴随着深度学习与机器...     

Data augmentation and parameter expansion for independent or spatially correlated ordinal data

Data augmentation and parameter expansion can lead to improved iterative sampling algorithms for Markov chain Monte Carlo (MCMC). Data augmentation allows for simpler and more feasible simulation from a posterior distribution. Parameter expansion accelerates convergence of iterative sampling algorithms by increasing the parameter space. Data augmentation and parameter-expanded data augmentation MCMC algorithms are proposed for fitting probit models for independent ordinal response data. The algorithms are extended for fitting probit linear mixed models for spatially correlated ordinal data. The effectiveness of data augmentation and parameter-expanded data augmentation is illustrated using the probit model and ordinal response data, however, the approach can be used broadly across model and data types.     

医学影像人工智能辅助诊断的样本增广方法

针对不同领域人工智能（AI）应用研究所面临的采用常规手段获取大量样本时耗时耗力耗财的问题,许多AI研究领域提出了各种各样的样本增广方法。首先,对样本增广的研究背景与意义进行介绍;其次,归纳了几种公知领域（包括自然图像识别、字符识别、语义分析）的样本增广方法,并在此基础上详细论述了医学影像辅助诊断方面的样本获取或增广方法,包括X光片、计算机断层成像（CT）图像、磁共振成像（MRI）图像的样本增广方法;最后,对AI应用领域数据增广方法存在的关键问题进行总结,并对未来的发展趋势进行展望。经归纳总结可知,获取足够数量且具有广泛代表性的训练样本是所有领域AI研发的关键环节。无论是公知领域还是专业领域都进行样本增广,且不同领域甚至同一领域的不同研究方向,其样本获取或增广方法均不相同。此外,样本增广并不是简单地增加样本数量,而是尽可能再现小样本量无法完全覆盖的真实样本存在,进而提高样本多样性,增强AI系统性能。     

基于自引导进化策略的高效自动化数据增强算法

深度学习在图像、文本、语音等媒体数据的分析任务上取得了优异的性能. 数据增强可以非常有效地提升训练数据的规模以及多样性, 从而提高模型的泛化性. 但是, 对于给定数据集, 设计优异的数据增强策略大量依赖专家经验和领域知识, 而且需要反复尝试, 费时费力. 近年来, 自动化数据增强通过机器自动设计数据增强策略, 已引起了学界和业界的广泛关注. 为了解决现有自动化数据增强算法尚无法在预测准确率和搜索效率之间取得良好平衡的问题, 提出一种基于自引导进化策略的自动化数据增强算法SGES AA. 首先, 设计一种有效的数据增强策略连续化向量表示方法, 并将自动化数据增强问题转换为连续化策略向量的搜索问题. 其次, 提出一种基于自引导进化策略的策略向量搜索方法, 通过引入历史估计梯度信息指导探索点的采样与更新, 在能够有效避免陷入局部最优解的同时, 可提升搜索过程的收敛速度. 在图像、文本以及语音数据集上的大量实验结果表明, 所提算法在不显著增加搜索耗时的情况下, 预测准确率优于或者匹配目前最优的自动化数据增强方法.     

深度学习图像数据增广方法研究综述

数据作为深度学习的驱动力，对于模型的训练至关重要。充足的训练数据不仅可以缓解模型在训练时的过拟合问题，而且可以进一步扩大参数搜索空间，帮助模型进一步朝着全局最优解优化。然而，在许多领域或任务中，获取到充足训练样本的难度和代价非常高。因此，数据增广成为一种常用的增加训练样本的手段。本文对目前深度学习中的图像数据增广方法进行研究综述，梳理了目前深度学习领域为缓解模型过拟合问题而提出的各类数据增广方法，按照方法本质原理的不同，将其分为单数据变形、多数据混合、学习数据分布和学习增广策略等4类方法，并以图像数据为主要研究对象，对各类算法进一步按照核心思想进行细分，并对方法的原理、适用场景和优缺点进行比较和分析，帮助研究者根据数据的特点选用合适的数据增广方法，为后续国内外研究者应用和发展研究数据增广方法提供基础。针对图像的数据增广方法，单数据变形方法主要可以分为几何变换、色域变换、清晰度变换、噪声注入和局部擦除等5种；多数据混合可按照图像维度的混合和特征空间下的混合进行划分；学习数据分布的方法主要基于生成对抗网络和图像风格迁移的应用进行划分；学习增广策略的典型方法则可以按照基于元学习和基于强化学习进行分类。目前，数据增广已然成为推进深度学习在各领域应用的一项重要技术，可以很有效地缓解训练数据不足带来的深度学习模型过拟合的问题，进一步提高模型的精度。在实际应用中可根据数据和任务的特点选择和组合最合适的方法，形成一套有效的数据增广方案，进而为深度学习方法的应用提供更强的动力。在未来，根据数据和任务基于强化学习探索最优的组合策略，基于元学习自适应地学习最优数据变形和混合方式，基于生成对抗网络进一步拟合真实数据分布以采样高质量的未知数据，基于风格迁移探索多模态数据互相转换的应用，这些研究方向十分值得探索并且具有广阔的发展前景。     

Augmenting Edge-Connectivity between Vertex Subsets

Given a directed or undirected graph G=(V,E), a collection ${\mathcal{R}}=\{(S_{i},T_{i}) \mid i=1,2,\ldots,|{\mathcal{R}}|, S_{i},T_{i} \subseteq V, S_{i} \cap T_{i} =\emptyset\}$ of two disjoint subsets of V, and a requirement function $r: {\mathcal{R}} \to\mathbb{R}_{+}$ , we consider the problem (called area-to-area edge-connectivity augmentation problem) of augmenting G by a smallest number of new edges so that the resulting graph $\hat{G}$ satisfies $d_{\hat{G}}(X)\geq r(S,T)$ for all X?V, $(S,T) \in{\mathcal{R}}$ with S?X?V?T, where d _G (X) denotes the degree of a vertex set X in G. This problem can be regarded as a natural generalization of the global, local, and node-to-area edge-connectivity augmentation problems. In this paper, we show that there exists a constant c such that the problem is inapproximable within a ratio of $c\log{|{\mathcal{R}}|}$ , unless P=NP, even restricted to the directed global node-to-area edge-connectivity augmentation or undirected local node-to-area edge-connectivity augmentation. We also provide an ${\mathrm{O}}(\log{|{\mathcal{R}}|})$ -approximation algorithm for the area-to-area edge-connectivity augmentation problem, which is a natural extension of Kortsarz and Nutov’s algorithm (Kortsarz and Nutov, J. Comput. Syst. Sci., 74:662–670, 2008). This together with the negative result implies that the problem is ${\varTheta}(\log{|{\mathcal{R}}|})$ -approximable, unless P=NP, which solves open problems for node-to-area edge-connectivity augmentation in Ishii et al. (Algorithmica, 56:413–436, 2010), Ishii and Hagiwara (Discrete Appl. Math., 154:2307–2329, 2006), Miwa and Ito (J. Oper. Res. Soc. Jpn., 47:224–243, 2004). Furthermore, we characterize the node-to-area and area-to-area edge-connectivity augmentation problems as the augmentation problems with modulotone and k-modulotone functions.     

A comprehensive survey of recent trends in deep learning for digital images augmentation

Deep learning proved its efficiency in many fields of computer science such as computer vision, image classifications, object detection, image segmentation, and more. Deep learning models primarily depend on the availability of huge datasets. Without the existence of many images in datasets, different deep learning models will not be able to learn and produce accurate models. Unfortunately, several fields don't have access to large amounts of evidence, such as medical image processing. For example. The world is suffering from the lack of COVID-19 virus datasets, and there is no benchmark dataset from the beginning of 2020. This pandemic was the main motivation of this survey to deliver and discuss the current image data augmentation techniques which can be used to increase the number of images. In this paper, a survey of data augmentation for digital images in deep learning will be presented. The study begins and with the introduction section, which reflects the importance of data augmentation in general. The classical image data augmentation taxonomy and photometric transformation will be presented in the second section. The third section will illustrate the deep learning image data augmentation. Finally, the fourth section will survey the state of the art of using image data augmentation techniques in the different deep learning research and application.

     

AVScreen: a real-time video augmentation method

Journal of Real-Time Image Processing - We present a tool for video augmentation in real-time, which we name the augmentation virtual screen (AVScreen). AVScreen is useful for developing...     

集成数据增强提高自动简答题评分的泛化性

自动简答题评分(Automated short answer grading,ASAG)是利用自然语言处理技术减少教育工作者人工评分负担。值得注意的是,目前大多数ASAG系统存在缺陷,学生通过复制或稍微改写标准答案取得高分的欺骗行为。该文探索一种基于规则的数据增强方法研究ASAG系统的鲁棒性。然而,由于自然语言存在离散性因素,导致基于规则的数据增强合成的样本的多样性受到限制。该文提出基于知识蒸馏的数据增强策略,以并行的方式堆叠不同的单个数据增强方法。此外,该文提出基于监督对比学习的ASAG系统,使得模型能学习到有效的句子表示。该文在University of North Texas和SemEval-2013两个公开数据集上进行了评估,与基线模型相比,该文提出的系统在性能上有实质性提高。     

Augmented Industrial Data-Driven Modeling Under the Curse of Dimensionality

The curse of dimensionality refers to the problem o increased sparsity and computational complexity when dealing with high-dimensional data. In recent years, the types and vari ables of industrial data have increased significantly, making data driven models more challenging to develop. To address this prob lem, data augmentation technology has been introduced as an effective tool to solve the sparsity problem of high-dimensiona industrial data. This paper systematically explores and discusses th...     

基于混合样本自动数据增强技术的半监督学习方法

基于一致性的半监督学习方法通常使用简单的数据增强方法来实现对原始输入和扰动输入的一致性预测.在有标签数据的比例较低的情况下,该方法的效果难以得到保证.将监督学习中一些先进的数据增强方法扩展到半监督学习环境中,是解决该问题的思路之一.基于一致性的半监督学习方法MixMatch,提出了基于混合样本自动数据增强技术的半监督学...     

多传感器时滞系统信息融合最优Kalman滤波器

基于线性最小方差最优加权融合估计算法,对多传感器的离散线性状态时滞随机系统,给出了一种非增广分布式加权融合最优Kalman滤波器.推导了状态时滞系统任两个传感器子系统之间的滤波误差互协方差阵的计算公式.它与状态增广加权融合滤波器具有相同的精度.与每个传感器的局部滤波器相比,分布式融合滤波器具有更高的精度.与状态和观测增广最优滤波器相比,具有较小的精度.但避免了增广所带来的高维计算和大的空间存储,可减小计算负担.仿真例子验证了其有效性.     

Calibration, Registration, and Synchronization for High Precision Augmented Reality Haptics

In our current research we examine the application of visuo-haptic augmented reality setups in medical training. To this end, highly accurate calibration, system stability, and low latency are indispensable prerequisites. These are necessary to maintain user immersion and avoid breaks in presence which potentially diminish the training outcome. In this paper we describe the developed calibration methods for visuo-haptic integration, the hybrid tracking technique for stable alignment of the augmentation, and the distributed framework ensuring low latency and component synchronization. Finally, we outline an early prototype system based on the multimodal augmented reality framework. The latter allows colocated visuo-haptic interaction with real and virtual scene components in a simplified open surgery setting.     

Novel soft bending actuator-based power augmentation hand exoskeleton controlled by human intention

This article presents the development of a soft material power augmentation wearable robot using novel bending soft artificial muscles. This soft exoskeleton was developed as a human hand power augmentation system for healthy or partially hand disabled individuals. The proposed prototype serves healthy manual workers by decreasing the muscular effort needed for grasping objects. Furthermore, it is a power augmentation wearable robot for partially hand disabled or post-stroke patients, supporting and augmenting the fingers’ grasping force with minimum muscular effort in most everyday activities. This wearable robot can fit any adult hand size without the need for any mechanical system changes or calibration. Novel bending soft actuators are developed to actuate this power augmentation device. The performance of these actuators has been experimentally assessed. A geometrical kinematic analysis and mathematical output force model have been developed for the novel actuators. The performance of this mathematical model has been proven experimentally with promising results. The control system of this exoskeleton is created by hybridization between cascaded position and force closed-loop intelligent controllers. The cascaded position controller is designed for the bending actuators to follow the fingers in their bending movements. The force controller is developed to control the grasping force augmentation. The operation of the control system with the exoskeleton has been experimentally validated. EMG signals were monitored during the experiments to determine that the proposed exoskeleton system decreased the muscular efforts of the wearer.     

面向中文命名实体识别任务的数据增强

In low-resource natural language processing (NLP) tasks, the existing data is not enough to train an ideal deep learning model. Text data augmentation is an effective method to improve the training effect of such tasks. This paper proposes a group of data augmentation methods based on instance substitution for the task of Chinese named entity recognition. A named entity in the training sample can be replaced by another entity of the same kind without changing the label. The specific algorithms include: 1) crossover substitution between existing entities; 2) synonymous replacement of entity components; 3) automatic generation of Chinese names. These methods are applied to PeopleDailyNER and CLUENER2020 datasets respectively, and the augmentation data is used to train the BERT+CRF model. The experimental results show that the F1 value of the model can be improved by about 10% and 7% respectively on the two datasets with only adding the same amount of augmentation data as the original data under the condition of small samples, and it also has a significant improvement when the training samples increase.     

Two augmentation preconditioners for nonsymmetric and indefinite saddle point linear systems with singular (1, 1) blocks

In this paper, two preconditioners based on augmentation are introduced for the solution of large saddle point-type systems with singular (1, 1) blocks. We study the spectral characteristics of the preconditioners, show that all eigenvalues of preconditioned matrices are strongly clustered. Finally, numerical experiments are also reported for illustrating the efficiency of the presented preconditioners.     

模糊联想记忆网络的增强学习算法

针对 Kosko提出的最大最小模糊联想记忆网络存在的问题 ,通过对这种网络连接权学习规则的改进 ,给出了另一种权重学习规则 ,即把 Kosko的前馈模糊联想记忆模型发展成为模糊双向联想记忆模型 ,并由此给出了模糊快速增强学习算法 ,该算法能存储任意给定的多值训练模式对集 .其中对于存储二值模式对集 ,由于其连接权值取值 0或 1,因而该算法易于硬件电路和光学实现 .实验结果表明 ,模糊快速增强学习算法是行之有效的 .     

MaskMix:用于变化检测的掩码混合数据增强方法

数据增强是提升变化检测模型泛化能力的一种主要方法。尽管现有的数据增强方法在图像分类、目标检测中取得了较好的效果,但忽略了多个时间序列图像之间的差异和变化目标的多样性。为了较好地保留变化区域并且增加复杂的背景信息,基于变化区域掩码,提出一种适用于变化检测的数据增强方法：MaskMix。首先,将当前图像对的变化区域粘贴到一个图像对上,得到具有新的背景和新的变化的变化图像对。其次,采用多路径加权融合策略进一步增强变化图像对。在每条路径上,从图像处理集合中随机选取一种经典的图像处理方法进一步处理变化图像对,然后使用Dirichlet分布产生的K维权重将K条路径处理后的图像对进行融合。最后,通过跳跃连接将处理前的图像对和处理后的图像对按Beta分布产生权重进行更深层次的混合。实验结果表明,提出的MaskMix在BCD和LEVIR-CD两个数据集上,有效地提升了变化检测方法ADCDNet、BIT、ChangeFormer、SNUNet和DSAMNet的泛化性能。与现有的图像增强方法MixUp、AugMix、MUM和CropMix相比,MaskMix能有效增加变化图像的复杂性和多样性,提升现有变化检测方法的泛化性能。