Medical diagnosis with C4.5 rule preceded by artificial neural network ensemble

Comprehensibility is very important when machine learning techniques are used in computer-aided medical diagnosis. Since an artificial neural network ensemble is composed of multiple artificial neural networks, its comprehensibility is worse than that of a single artificial neural network. In this paper, C4.5 Rule-PANE, which combines an artificial neural network ensemble with rule induction by regarding the former as a preprocess of the latter, is proposed. At first, an artificial neural network ensemble is trained. Then, a new training data set is generated by feeding the feature vectors of original training instances to the trained ensemble and replacing the expected class labels of original training instances with the class labels output from the ensemble. Additional training data may also be appended by randomly generating feature vectors and combining them with their corresponding class labels output from the ensemble. Finally, a specific rule induction approach, i.e., C4.5 Rule, is used to learn rules from the new training data set. Case studies on diabetes, hepatitis , and breast cancer show that C4.5 Rule-PANE could generate rules with strong generalization ability, which benefits from an artificial neural network ensemble, and strong comprehensibility, which benefits from rule induction.     

基于U2-Net和VGG16的天然气泄露检测

目前在深度学习领域很少以天然气泄露图像为数据进行研究,本文使用甲烷红外图像训练的卷积神经网络(VGG16)来实现泄露检测。另外,针对泄露的甲烷气体与背景图像存在相似性的问题,使用U²-Net图像分割网络代替背景建模方法来提取泄露气体区域。通过迁移VGG16网络模型结构和卷积层参数,在卷积层和激励层之间加入BN层以提高训练速度,将最后一层池化层替换为基于最大池化算法的动态自适应池化方法以提高检测精度。将改进的VGG16神经网络对分割的红外图像进行训练并与其他卷积神经网络进行对比,使用准确率,精准率,召回率和F1-score来对模型进行综合评价,其表现效果最好。与现有的检测方法进行对比,所提出的检测方法准确率更高。该检测方法能够实现高精度泄漏检测,满足天然气泄露检测准确性的要求,且模型具有较好的泛化能力和鲁棒性。     

Knowledge-Based Artificial Neural Network Models for Microstrip Radial Stubs

Radial stubs are a superior choice over low characteristic impedance rectangular stubs in terms of providing an accurate localized zero-impedance reference point and maintaining a low input impedance value over a wide frequency range. In this paper, knowledge-based artificial neural networks are used to model the microstrip radial stubs. Using space-mapping technology and Huber optimization make the neural network models for radial stubs decrease the number of training data, improve generalization ability, and reduce the complexity of the neural network topology with respect to the classical neuromodeling approach. The neural networks are developed for design and optimization of radial stubs, which are robust both from the angle of time of computation and accuracy.     

BP网络学习能力与泛化能力之间的定量关系式

分析BP网络过拟合时网络学习能力与泛化能力之间的内在联系,引入描述问题复杂性程度的复相关系数,建立了BP网络过拟合时,反映网络学习能力的训练样本集的训练相对误差与表征泛化能力的网络对检验样本集的测试相对误差之间满足的定量关系式.通过模拟若干不同类型函数的BP网络数值建模试验,确定了关系式中过拟合参数q的取值范围为0.007~0.07,指出BP网络应用于给定样本集的训练过程中,具有较佳泛化能力的停止训练方法.     

Fast Classification Networks For Signal Processing

We present a generalization of the corner classification approach to training feedforward neural networks that allows rapid learning of nonbinary data. These generalized networks, called fast classification (FC) networks, are compared against backpropagation and radial basis function networks and are shown to have excellent performance for prediction of time series and pattern recognition. FC networks do not require iterative training and they can be used in many signal processing applications where fast, nonlinear filtering provides an advantage.     

Neural networks for vector quantization of speech and images

Using neural networks for vector quantization (VQ) is described. The authors show how a collection of neural units can be used efficiently for VQ encoding, with the units performing the bulk of the computation in parallel, and describe two unsupervised neural network learning algorithms for training the vector quantizer. A powerful feature of the new training algorithms is that the VQ codewords are determined in an adaptive manner, compared to the popular LBG training algorithm, which requires that all the training data be processed in a batch mode. The neural network approach allows for the possibility of training the vector quantizer online, thus adapting to the changing statistics of the input data. The authors compare the neural network VQ algorithms to the LBG algorithm for encoding a large database of speech signals and for encoding images     

红外目标检测网络改进半监督迁移学习方法

针对红外数据集规模小,标记样本少的特点,提出了一种红外目标检测网络的半监督迁移学习方法,主要用于提高目标检测网络在小样本红外数据集上的训练效率和泛化能力,提高深度学习模型在训练样本较少的红外目标检测等场景当中的适应性。文中首先阐述了在标注样本较少时无标注样本对提高模型泛化能力、抑制过拟合方面的作用。然后提出了红外目标检测网络的半监督迁移学习流程:在大量的RGB图像数据集中训练预训练模型,后使用少量的有标注红外图像和无标注红外图像对网络进行半监督学习调优。另外,文中提出了一种特征相似度加权的伪监督损失函数,使用同一批次样本的预测结果相互作为标注,以充分利用无标注图像内相似目标的特征分布信息;为降低半监督训练的计算量,在伪监督损失函数的计算中,各目标仅将其特征向量邻域范围内的预测目标作为伪标注。实验结果表明,文中方法所训练的目标检测网络的测试准确率高于监督迁移学习所获得的网络,其在Faster R-CNN上实现了1.1%的提升,而在YOLO-v3上实现了4.8%的显著提升,验证了所提出方法的有效性。     

Meta-transfer-adjustment learning for few-shot learning

Deep neural network models with strong feature extraction capacity are prone to overfitting and fail to adapt quickly to new tasks with few samples. Gradient-based meta-learning approaches can minimize overfitting and adapt to new tasks fast, but they frequently use shallow neural networks with limited feature extraction capacity. We present a simple and effective approach called Meta-Transfer-Adjustment learning (MTA) in this paper, which enables deep neural networks with powerful feature extraction capabilities to be applied to few-shot scenarios while avoiding overfitting and gaining the capacity for quickly adapting to new tasks via training on numerous tasks. Our presented approach is classified into two major parts, the Feature Adjustment (FA) module, and the Task Adjustment (TA) module. The feature adjustment module (FA) helps the model to make better use of the deep network to improve feature extraction, while the task adjustment module (TA) is utilized for further improve the model’s fast response and generalization capabilities. The proposed model delivers good classification results on the benchmark small sample datasets MiniImageNet and Fewshot-CIFAR100, as proved experimentally.     

一种全新的两步自动化医学图像分割方案

针对现有图像分割方法存在需要手动分割,以及精确度较低的问题。采用一种全新的两步图像分割方案。该方案。以基于人工神经网络的模式识别技术,即人工神经网络的大规模培训的方法,通过对肺区不同子区域内结构进行分割处理,利用训练好的大规模人工神经网络对标准胸片中的肋骨、锁骨等骨质结构进行抑制,结合以基于区域的活动轮廓模型,即Snake模型,正确分割亮度不均匀的图像。文中选择与医护人员人工分割的图像进行对比,通过放射科医生采用等级法打分,原图的平均分为20分,而通过文中改进的分割方法平均分高达34分。     

Semi-supervised cardiac MRI image of the left ventricle segmentation algorithm based on contrastive learning

A semi-supervised convolutional neural network segmentation method of medical images based on contrastive learning is proposed. The cardiac magnetic resonance imaging(MRI) images to be segmented are preprocessed to obtain positive and negative samples by labels. The U-Net shrinks network is applied to extract features of the positive samples, negative samples, and input samples. In addition, an unbalanced contrastive loss function is proposed, which is weighted with the binary cross-entropy loss...     

Conditional distribution learning with neural networks and itsapplication to channel equalization

We present a conditional distribution learning formulation for real-time signal processing with neural networks based on an extension of maximum likelihood theory-partial likelihood (PL) estimation-which allows for (i) dependent observations and (ii) sequential processing. For a general neural network conditional distribution model, we establish a fundamental information-theoretic connection, the equivalence of maximum PL estimation, and accumulated relative entropy (ARE) minimization, and obtain large sample properties of PL for the general case of dependent observations. As an example, the binary case with the sigmoidal perceptron as the probability model is presented. It is shown that the single and multilayer perceptron (MLP) models satisfy conditions for the equivalence of the two cost functions: ARE and negative log partial likelihood. The practical issue of their gradient descent minimization is then studied within the well-formed cost functions framework. It is shown that these are well-formed cost functions for networks without hidden units; hence, their gradient descent minimization is guaranteed to converge to a solution if one exists on such networks. The formulation is applied to adaptive channel equalization, and simulation results are presented to show the ability of the least relative entropy equalizer to realize complex decision boundaries and to recover during training from convergence at the wrong extreme in cases where the mean square error-based MLP equalizer cannot     

Gradient-based learning applied to document recognition

Multilayer neural networks trained with the back-propagation algorithm constitute the best example of a successful gradient based learning technique. Given an appropriate network architecture, gradient-based learning algorithms can be used to synthesize a complex decision surface that can classify high-dimensional patterns, such as handwritten characters, with minimal preprocessing. This paper reviews various methods applied to handwritten character recognition and compares them on a standard handwritten digit recognition task. Convolutional neural networks, which are specifically designed to deal with the variability of 2D shapes, are shown to outperform all other techniques. Real-life document recognition systems are composed of multiple modules including field extraction, segmentation recognition, and language modeling. A new learning paradigm, called graph transformer networks (GTN), allows such multimodule systems to be trained globally using gradient-based methods so as to minimize an overall performance measure. Two systems for online handwriting recognition are described. Experiments demonstrate the advantage of global training, and the flexibility of graph transformer networks. A graph transformer network for reading a bank cheque is also described. It uses convolutional neural network character recognizers combined with global training techniques to provide record accuracy on business and personal cheques. It is deployed commercially and reads several million cheques per day     

基于深度学习的语义分割网络

提出了一种基于深度学习的语义分割网络。该网络通过多孔卷积设计了一个能提取图像多尺度信息的空间金字塔模块,并通过大量实验探索了空间金字塔模块中多孔采样率和多尺度分支对于网络场景解析能力的影响。讨论了网络训练中不同超参数对于网络性能的影响。在SUN RGB-D数据集上的测试结果显示,与其它state-of-the-art的语义分割网络相比,本文设计的网络性能突出。最后,还对基于红外图像的语义分割进行了初步探索。     

Automated segmentation and classification of high throughput yeast assay spots

Several technologies for characterizing genes and proteins from humans and other organisms use yeast growth or color development as read outs. The yeast two-hybrid assay, for example, detects protein-protein interactions by measuring the growth of yeast on a specific solid medium, or the ability of the yeast to change color when grown on a medium containing a chromogenic substrate. Current systems for analyzing the results of these types of assays rely on subjective and inefficient scoring of growth or color by human experts. Here, an image analysis system is described for scoring yeast growth and color development in high throughput biological assays. The goal is to locate the spots and score them in color images of two types of plates named "X-Gal" and "growth assay" plates, with uniformly placed spots (cell areas) on each plate (both plates in one image). The scoring system relies on color for the X-Gal spots, and texture properties for the growth assay spots. A maximum likelihood projection-based segmentation is developed to automatically locate spots of yeast on each plate. Then color histogram and wavelet texture features are extracted for scoring using an optimal linear transformation. Finally, an artificial neural network is used to score the X-Gal and growth assay spots using the extracted features. The performance of the system is evaluated using spots of 60 images. After training the networks using training and validation sets, the system was assessed on the test set. The overall accuracies of 95.4% and 88.2% are achieved, respectively, for scoring the X-Gal and growth assay spots.     

用无需选取参数的Unit-linking PCNN进行自动图像分割

脉冲耦合神经网络（PCNN-Pulse Coupled Neural Network）是一种有生物学依据的人工神经网络，它可有效地用于图像分割。基于PCNN的图像分割效果取决于PCNN中各参数的选择。然而，图像分割时，各种不同的图像对应的PCNN参数是不同的，而PCNN参数的选择是困难的。本文提出了一种基于Unit-linking PCNN的图像分割新方法，解决了PCNN图像分割参数选择的难题。用本文提出的新方法可有效地自动分割各种图像，而无需考虑PCNN参数的选择，这对于PCNN的理论研究和实际应用有重要的意义。     

一种投票权值调整的神经网络集成增量学习方法

针对神经网络集成增量学习中集成输出投票权值的设定问题,给出了一种投票权值调整的神经网络集成增量学习方法。该方法定义了神经网络集成中子神经网络训练集的类核函数,通过计算待识样本与类核函数之间的核函数距离得到集成输出中子神经网络的投票权值。这种投票权值设定方法可以根据子神经网络分类器对待识样本的分类性能自适应地调整集成输出的投票权值,是一种更加合理的集成输出投票权值设定方法。仿真实验表明,这种投票权值调整的神经网络集成增量学习方法比投票权值固定的方法增量学习性能更优。      

Agglomerative Learning Algorithms for General Fuzzy Min-Max Neural Network

In this paper two agglomerative learning algorithms based on new similarity measures defined for hyperbox fuzzy sets are proposed. They are presented in a context of clustering and classification problems tackled using a general fuzzy min-max (GFMM) neural network. The proposed agglomerative schemes have shown robust behaviour in presence of noise and outliers and insensitivity to the order of training patterns presentation. The emphasis is also put on the complimentary features to the previously presented incremental learning scheme more suitable for on-line adaptation and dealing with large training data sets. The performance and other properties of the agglomerative schemes are illustrated using a number of artificial and real-world data sets.     

Multispectral classification of Landsat-images using neuralnetworks

The authors report the application of three-layer back-propagation networks for classification of Landsat TM data on a pixel-by-pixel basis. The results are compared to Gaussian maximum likelihood classification. First, it is shown that the neural network is able to perform better than the maximum likelihood classifier. Secondly, in an extension of the basic network architecture it is shown that textural information can be integrated into the neural network classifier without the explicit definition of a texture measure. Finally, the use of neural networks for postclassification smoothing is examined     

Neural network based adaptable control method for linearization of high power amplifiers

The proposed adaptable control method for linearization of high power amplifiers is powered by the neural networks technique that supports analogue polynomial type of predistorters, which are widely utilized in commercial power amplifiers for wireless communication purposes. This paper presents an algorithm to determine the coefficients by use of artificial neural networks and its generalization feature that helps to map the power amplifier response with optimal coefficients of the polynomial, which in a proper way pre-distorts an input signal of the amplifier. The concept of the predistortion has been introduced. Furthermore, the overall step-by-step initialization and functionality of the control process has also been described. The method has been tested successfully in a real power amplifier equipped with an analogue predistorting circuits. Presented measurement results imply that this approach is robust and well suited for such category of the power amplifier design, in which the artificial neural networks play substantial role.