基于深度学习的数字病理图像分割综述与展望

数字病理图像分析对于乳腺癌、前列腺癌等良恶性分级诊断具有重要意义,其中组织基元的形态和目标测量是量化分析的重要依据.然而,由于病理数据多样性和复杂性等新特点,其分割任务面临着特征提取困难、实例分割困难等挑战.人工智能辅助病理量化分析,将复杂病理数据转化为可挖掘的图像特征,使得自动提取组织基元的定量化信息成为可能.特别是随着计算机计算能力的快速发展,深度学习技术凭借其强大的特征学习、设计灵活等特性在数字病理量化分析领域取得了突破性成果.本文系统概述目前代表性深度学习方法,包括卷积神经网络、全卷积网络、编码器—解码器模型、循环神经网络、生成对抗网络等方法体系,总结深度学习在病理图像分割等任务中的建模机理和应用,并梳理了现有方法的方法理论、关键技术、优缺点和性能分析.最后,本文讨论了未来数字病理图像分割深度学习建模的开放性挑战和新趋势.     

Statistical topic models for multi-label document classification

Machine learning approaches to multi-label document classification have to date largely relied on discriminative modeling techniques such as support vector machines. A?drawback of these approaches is that performance rapidly drops off as the total number of labels and the number of labels per document increase. This problem is amplified when the label frequencies exhibit the type of highly skewed distributions that are often observed in real-world datasets. In this paper we investigate a class of generative statistical topic models for multi-label documents that associate individual word tokens with different labels. We investigate the advantages of this approach relative to discriminative models, particularly with respect to classification problems involving large numbers of relatively rare labels. We compare the performance of generative and discriminative approaches on document labeling tasks ranging from datasets with several thousand labels to datasets with tens of labels. The experimental results indicate that probabilistic generative models can achieve competitive multi-label classification performance compared to discriminative methods, and have advantages for datasets with many labels and skewed label frequencies.     

基于特征通道和空间位置注意力的三维点云特征学习网络

点云模型的分类与部件分割是三维点云数据处理的基本任务,其核心在于获取可以有效表示三维模型的点云特征。提出一个引入注意力机制的三维点云特征学习网络。该网络采用多层次点云特征提取方法,首先使用特征通道注意力模块获取各通道间的关联,增强关键通道信息; 接着引入空间位置注意力机制,基于点的空间位置信息获取各点的注意力权重;然后结合以上2种注意力机制获取增强的点云特征;最后基于该特征继续进行多层次特征提取,获得面向下游任务的点云特征。分别在ModelNet40和ShapeNet数据集上进行形状分类与部件分割实验,结果表明,使用所提方法可以实现高精度、具有鲁棒性的三维点云形状分类与分割。     

Cascaded model adaptation for dialog act segmentation and tagging

There are many speech and language processing problems which require cascaded classification tasks. While model adaptation has been shown to be useful in isolated speech and language processing tasks, it is not clear what constitutes system adaptation for such complex systems. This paper studies the following questions: In cases where a sequence of classification tasks is employed, how important is to adapt the earlier or latter systems? Is the performance improvement obtained in the earlier stages via adaptation carried on to later stages in cases where the later stages perform adaptation using similar data and/or methods? In this study, as part of a larger scale multiparty meeting understanding system, we analyze various methods for adapting dialog act segmentation and tagging models trained on conversational telephone speech (CTS) to meeting style conversations. We investigate the effect of using adapted and unadapted models for dialog act segmentation with those of tagging, showing the effect of model adaptation for cascaded classification tasks. Our results indicate that we can achieve significantly better dialog act segmentation and tagging by adapting the out-of-domain models, especially when the amount of in-domain data is limited. Experimental results show that it is more effective to adapt the models in the latter classification tasks, in our case dialog act tagging, when dealing with a sequence of cascaded classification tasks.     

Transfer Learning from Unlabeled Data via Neural Networks

A machine learning framework which uses unlabeled data from a related task domain in supervised classification tasks is described. The unlabeled data come from related domains, which share the same class labels or generative distribution as the labeled data. Patterns in the unlabeled data are learned via a neural network and transferred to the target domain from where the labeled data are generated, so as to improve the performance of the supervised learning task. We call this approach self-taught transfer learning from unlabeled data. We introduce a general-purpose feature learning algorithm producing features that retain information from the unlabeled data. Information preservation assures that the features obtained will be useful for improving the classification performance of the supervised tasks.     

Large margin mixture of AR models for time series classification

In this paper, we propose the large margin autoregressive (LMAR) model for classification of time series patterns. The parameters of the generative AR models for different classes are estimated using the margin of the boundaries of AR models as the optimization criterion. Models that use a mixture of AR (MAR) models are considered for representing the data that cannot be adequately represented using a single AR model for a class. Based on a mixture model representing each class, we propose the large margin mixture of AR (LMMAR) models. The proposed methods are applied on the simulated time series data, electrocardiogram data, speech data for E-set in English alphabet and electroencephalogram time series data. Performance of the proposed methods is compared with that of support vector machine (SVM) based classifier that uses AR coefficients based features. The proposed methods give a better classification performance compared to the SVM based classifier. Being generative models, the LMAR and LMMAR models provide a generative interpretation that enables utilization of the rejection option in the high risk classification tasks. The proposed methods can also be used for detection of novel time series data.     

Boosted Bayesian network classifiers

The use of Bayesian networks for classification problems has received a significant amount of recent attention. Although computationally efficient, the standard maximum likelihood learning method tends to be suboptimal due to the mismatch between its optimization criteria (data likelihood) and the actual goal of classification (label prediction accuracy). Recent approaches to optimizing classification performance during parameter or structure learning show promise, but lack the favorable computational properties of maximum likelihood learning. In this paper we present boosted Bayesian network classifiers, a framework to combine discriminative data-weighting with generative training of intermediate models. We show that boosted Bayesian network classifiers encompass the basic generative models in isolation, but improve their classification performance when the model structure is suboptimal. We also demonstrate that structure learning is beneficial in the construction of boosted Bayesian network classifiers. On a large suite of benchmark data-sets, this approach outperforms generative graphical models such as naive Bayes and TAN in classification accuracy. Boosted Bayesian network classifiers have comparable or better performance in comparison to other discriminatively trained graphical models including ELR and BNC. Furthermore, boosted Bayesian networks require significantly less training time than the ELR and BNC algorithms.     

Hybrid Generative-Discriminative Visual Categorization

Learning models for detecting and classifying object categories is a challenging problem in machine vision. While discriminative approaches to learning and classification have, in principle, superior performance, generative approaches provide many useful features, one of which is the ability to naturally establish explicit correspondence between model components and scene features—this, in turn, allows for the handling of missing data and unsupervised learning in clutter. We explore a hybrid generative/discriminative approach, using ‘Fisher Kernels’ (Jaakola, T., et al. in Advances in neural information processing systems, Vol. 11, pp. 487–493, 1999), which retains most of the desirable properties of generative methods, while increasing the classification performance through a discriminative setting. Our experiments, conducted on a number of popular benchmarks, show strong performance improvements over the corresponding generative approach. In addition, we demonstrate how this hybrid learning paradigm can be extended to address several outstanding challenges within computer vision including how to combine multiple object models and learning with unlabeled data.     

Leveraging Graph Cut’s Energy Function for Context Aware Facial Recognition in Indoor Environments

Context-aware facial recognition regards the recognition of faces in association with their respective environments. This concept is useful for the domestic robot which interacts with humans when performing specific functions in indoor environments. Deep learning models have been relevant in solving facial and place recognition challenges; however, they require the procurement of training images for optimal performance. Pre-trained models have also been offered to reduce training time significantly. Regardless, for classification tasks, custom data must be acquired to ensure that learning models are developed from other pre-trained models. This paper proposes a place recognition model that is inspired by the graph cut energy function, which is specifically designed for image segmentation. Common objects in the considered environment are identified and thereafter they are passed over to a graph cut inspired model for indoor environment classification. Additionally, faces in the considered environment are extracted and recognised. Finally, the developed model can recognise a face together with its environment. The strength of the proposed model lies in its ability to classify indoor environments without the usual training process(es). This approach differs from what is obtained in traditional deep learning models. The classification capability of the developed model was compared to state-of-the-art models and exhibited promising outcomes.     

基于度量学习的无监督域适应方法及其在死亡风险预测上的应用

近年来,深度学习模型已在医疗领域的预测任务上得到广泛应用,并取得了不错的效果.然而,深度学习模型常会面临带标签训练数据不足、整体数据分布偏移和类别之间数据分布偏移的问题,导致模型预测的准确度下降.为解决上述问题,提出一种基于域对抗和加性余弦间隔损失的无监督域适应方法(additive margin softmax ba...     

基于多任务学习的生成式阅读理解

生成式阅读理解是机器阅读理解领域一项新颖且极具挑战性的研究。与主流的抽取式阅读理解相比,生成式阅读理解模型不再局限于从段落中抽取答案,而是能结合问题和段落生成自然和完整的表述作为答案。然而,现有的生成式阅读理解模型缺乏对答案在段落中的边界信息以及对问题类型信息的理解。为解决上述问题,该文提出一种基于多任务学习的生成式阅读理解模型。该模型在训练阶段将答案生成任务作为主任务,答案抽取和问题分类任务作为辅助任务进行多任务学习,同时学习和优化模型编码层参数;在测试阶段加载模型编码层进行解码生成答案。实验结果表明,答案抽取模型和问题分类模型能够有效提升生成式阅读理解模型的性能。     

Minimum classification error learning for sequential data in the wavelet domain

Wavelet analysis has found widespread use in signal processing and many classification tasks. Nevertheless, its use in dynamic pattern recognition have been much more restricted since most of wavelet models cannot handle variable length sequences properly. Recently, composite hidden Markov models which observe structured data in the wavelet domain were proposed to deal with this kind of sequences. In these models, hidden Markov trees account for local dynamics in a multiresolution framework, while standard hidden Markov models capture longer correlations in time. Despite these models have shown promising results in simple applications, only generative approaches have been used so far for parameter estimation. The goal of this work is to take a step forward in the development of dynamic pattern recognizers using wavelet features by introducing a new discriminative training method for this Markov models. The learning strategy relies on the minimum classification error approach and provides re-estimation formulas for fully non-tied models. Numerical experiments on phoneme recognition show important improvement over the recognition rate achieved by the same models trained using maximum likelihood estimation.     

生成对抗网络在医学图像处理中的应用

生成对抗网络（Generative Adversarial Nets,GANs）模型可以无监督学习到更丰富的数据信息,其包括生成模型与判别模型,凭借二者之间的对抗提高性能。针对传统GANs存在着梯度消失、模式崩溃及无法生成离散数据分布等问题,已经提出了大量的变体模型。介绍了生成对抗网络模型的理论和组成结构;介绍了几种典型的变体模型,重点介绍了生成对抗网络模型在图像生成、图像分割、图像分类、目标检测及图像超分辨率重建应用领域上的研究进展及现状。在研究现状和问题基础上进行了深入分析,进一步总结和探讨了GANs模型在医学图像处理领域中未来发展的趋势和所面临的挑战。     

视觉Transformer识别任务研究综述

Transformer模型在自然语言处理领域取得了很好的效果,同时因其能够更好地连接视觉和语言,也激发了计算机视觉界的极大兴趣。本文总结了视觉Transformer处理多种识别任务的百余种代表性方法,并对比分析了不同任务内的模型表现,在此基础上总结了每类任务模型的优点、不足以及面临的挑战。根据识别粒度的不同,分别着眼于诸如图像分类、视频分类的基于全局识别的方法,以及目标检测、视觉分割的基于局部识别的方法。考虑到现有方法在3种具体识别任务的广泛流行,总结了在人脸识别、动作识别和姿态估计中的方法。同时,也总结了可用于多种视觉任务或领域无关的通用方法的研究现状。基于Transformer的模型实现了许多端到端的方法,并不断追求准确率与计算成本的平衡。全局识别任务下的Transformer模型对补丁序列切分和标记特征表示进行了探索,局部识别任务下的Transformer模型因能够更好地捕获全局信息而取得了较好的表现。在人脸识别和动作识别方面,注意力机制减少了特征表示的误差,可以处理丰富多样的特征。Transformer可以解决姿态估计中特征错位的问题,有利于改善基于回归的方法性能,还减少了三维估计时深度映射所产生的歧义。大量探索表明视觉Transformer在识别任务中的有效性,并且在特征表示或网络结构等方面的改进有利于提升性能。     

深度学习在医学影像中的应用综述

深度学习能自动从大样本数据中学习获得优良的特征表达,有效提升各种机器学习任务的性能,已广泛应用于信号处理、计算机视觉和自然语言处理等诸多领域。基于深度学习的医学影像智能计算是目前智慧医疗领域的研究热点,其中深度学习方法已经应用于医学影像处理、分析的全流程。由于医学影像内在的特殊性、复杂性,特别是考虑到医学影像领域普遍存在的小样本问题,相关学习任务和应用场景对深度学习方法提出了新要求。本文以临床常用的X射线、超声、计算机断层扫描和磁共振等4种影像为例,对深度学习在医学影像中的应用现状进行综述,特别面向图像重建、病灶检测、图像分割、图像配准和计算机辅助诊断这5大任务的主要深度学习方法的进展进行介绍,并对发展趋势进行展望。     

Deep learning techniques for tumor segmentation: a review

Recently, deep learning, especially convolutional neural networks, has achieved the remarkable results in natural image classification and segmentation. At the same time, in the field of medical image segmentation, researchers use deep learning techniques for tasks such as tumor segmentation, cell segmentation, and organ segmentation. Automatic tumor segmentation plays an important role in radiotherapy and clinical practice and is the basis for the implementation of follow-up treatment programs. This paper reviews the tumor segmentation methods based on deep learning in recent years. We first introduce the common medical image types and the evaluation criteria of segmentation results in tumor segmentation. Then, we review the tumor segmentation methods based on deep learning from technique view and tumor view, respectively. The technique view reviews the researches from the architecture of the deep learning and the tumor view reviews from the type of tumors.

     

Unsupervised Learning of Probabilistic Object Models (POMs) for Object Classification, Segmentation, and Recognition Using Knowledge Propagation

We present a method to learn probabilistic object models (POMs) with minimal supervision, which exploit different visual cues and perform tasks such as classification, segmentation, and recognition. We formulate this as a structure induction and learning task and our strategy is to learn and combine elementary POMs that make use of complementary image cues. We describe a novel structure induction procedure, which uses knowledge propagation to enable POMs to provide information to other POMs and “teach them” (which greatly reduces the amount of supervision required for training and speeds up the inference). In particular, we learn a POM-IP defined on Interest Points using weak supervision [1], [2] and use this to train a POM-mask, defined on regional features, which yields a combined POM that performs segmentation/localization. This combined model can be used to train POM-edgelets, defined on edgelets, which gives a full POM with improved performance on classification. We give detailed experimental analysis on large data sets for classification and segmentation with comparison to other methods. Inference takes five seconds while learning takes approximately four hours. In addition, we show that the full POM is invariant to scale and rotation of the object (for learning and inference) and can learn hybrid objects classes (i.e., when there are several objects and the identity of the object in each image is unknown). Finally, we show that POMs can be used to match between different objects of the same category, and hence, enable objects recognition.     

基于有监督Topic Model的图像分类

近年来,以LDA为代表的话题模型在图像和文本处理中均得到了广泛的应用。与传统的机器学习方法相比,LDA模型具有参数少,表达能力强等优点,同时作为一种生成模型,它可以有效模拟人类学习的方式,便利地加入先验知识。有监督的LDA模型则将生成模型与判别模型结合在一起,是一种通用的分类方法。Dense-SIFT特征被作为底层特征,在词袋模型的框架下,以k-means算法构建词典,用有监督的LDA模型训练,并在通用的图像数据集上进行评测,根据评测结果证明其在图像分类任务中具有很好的性能。     

基于元学习的小样本数据生成算法

小样本数据存在信息不充足、不完备等问题,缺乏对总体的代表性,导致数据驱动的相关算法精度下降.本文针对小样本问题,提出基于元学习的生成式对抗网络算法进行小样本数据的数据生成.该算法目标是在各种数据生成任务上训练,确定模型最优初始化参数,从而仅使用较少的训练样本解决新的数据生成任务.本文利用水冷磁悬浮机组数据进行数据生成,实验表明,本算法能够在样本不足的条件下确定最优初始化参数,降低了对数据集大小的要求.本文同时进行了真实数据与生成数据混合的故障分类实验,验证了生成数据具有较好的真实性,对故障诊断分析具有较大的帮助.     

深度学习在组织病理学中的应用综述

组织病理学是临床上肿瘤诊断的金标准,直接关系到治疗的开展与预后的评估。来自临床的需求为组织病理诊断提出了质量与效率两个方面的挑战。组织病理诊断涉及大量繁重的病理切片判读任务,高度依赖医生的经验,但病理医生的培养周期长,人才储备缺口巨大,病理科室普遍超负荷工作。近年来出现的基于深度学习的组织病理辅助诊断方法可以帮助医生提高诊断工作的精度与速度,缓解病理诊断资源不足的问题,引起了研究人员的广泛关注。本文初步综述深度学习方法在组织病理学中的相关研究工作。介绍了组织病理诊断的医学背景,整理了组织病理学领域的主要数据集,重点介绍倍受关注的乳腺癌、淋巴结转移癌、结肠癌的病理数据及其分析任务。本文归纳了数据的存储与处理、模型的设计与优化以及小样本与弱标注学习这3项需要解决的技术问题。围绕这些问题,本文介绍了包括数据存储、数据预处理、分类模型、分割模型、迁移学习和多示例学习等相关研究工作。最后总结了面向组织病理学诊断的深度学习方法研究现状,并指出当下研究工作可能的改进方向。