A novel image Denoising approach using super resolution densely connected convolutional networks

Multimedia Tools and Applications - Image distortion effects, called noise, may occur due to various reasons such as image acquisition, transfer, and duplication. Image denoising is a preliminary...     

Controlling activity fluctuations in large, sparsely connected random networks

Controlling activity in recurrent neural network models of brain regions is essential both to enable effective learning and to reproduce the low activities that exist in some cortical regions such as hippocampal region CA3. Previous studies of sparse, random, recurrent networks constructed with McCulloch-Pitts neurons used probabilistic arguments to set the parameters that control activity. Here, we extend this work by adding an additional, biologically appropriate, parameter to control the magnitude and stability of activity oscillations. The new constant can be considered to be the rest conductance in a shunting model or the threshold when subtractive inhibition is used. This new parameter is critical for large networks run at low activity levels. Importantly, extreme activity fluctuations that act to turn large networks totally on or totally off can now be avoided. We also show how the size of external input activity interacts with this parameter to affect network activity. Then the model based on fixed weights is extended to estimate activities in networks with distributed weights. Because the theory provides accurate control of activity fluctuations, the approach can be used to design a predictable amount of pseudorandomness into deterministic networks. Such nonminimal fluctuations improve learning in simulations trained on the transitive inference problem.     

Evolutionary optimization of sparsely connected and time-lagged neural networks for time series forecasting

Time series forecasting (TSF) is an important tool to support decision making (e.g., planning production resources). Artificial neural networks (ANNs) are innate candidates for TSF due to advantages such as nonlinear learning and noise tolerance. However, the search for the best model is a complex task that highly affects the forecasting performance. In this work, we propose two novel evolutionary artificial neural networks (EANNs) approaches for TSF based on an estimation distribution algorithm (EDA) search engine. The first new approach consist of sparsely connected evolutionary ANN (SEANN), which evolves more flexible ANN structures to perform multi-step ahead forecasts. The second one, consists of an automatic Time lag feature selection EANN (TEANN) approach that evolves not only ANN parameters (e.g., input and hidden nodes, training parameters) but also which set of time lags are fed into the forecasting model. Several experiments were held, using a set of six time series, from different real-world domains. Also, two error metrics (i.e., mean squared error and symmetric mean absolute percentage error) were analyzed. The two EANN approaches were compared against a base EANN (with no ANN structure or time lag optimization) and four other methods (autoregressive integrated moving average method, random forest, echo state network and support vector machine). Overall, the proposed SEANN and TEANN methods obtained the best forecasting results. Moreover, they favor simpler neural network models, thus requiring less computational effort when compared with the base EANN.     

基于密集连接网络的图像隐写分析

针对目前传统的隐写分析技术对特征集要求越来越高的问题,构建了一个密集连接网络模型（Steganalysis-Densely Connected Convolutional Networks,S-DCCN）进行图像隐写分析,避免了人工提取特征,提高了隐写分析效率。首先,在网络层之前添加高通滤波层（HPF）进行滤波,加快模型训练速度。经过滤波后的图像进入两层卷积层进行特征提取,在卷积层之后使用了5组密集连接模块来解决网络加深带来的梯度消失问题,密集连接模块之间通过过度层来控制整个网络的宽度。实验结果表明,相比传统的图像隐写分析算法和卷积神经网络技术,该模型有效提高了隐写分析的准确率和泛化性能。     

Noise injection into inputs in sparsely connected Hopfield andwinner-take-all neural networks

In this paper, we show that noise injection into inputs in unsupervised learning neural networks does not improve their performance as it does in supervised learning neural networks. Specifically, we show that training noise degrades the classification ability of a sparsely connected version of the Hopfield neural network, whereas the performance of a sparsely connected winner-take-all neural network does not depend on the injected training noise.     

Polynomial-based graph convolutional neural networks for graph classification

Machine Learning - Graph convolutional neural networks exploit convolution operators, based on some neighborhood aggregating scheme, to compute representations of graphs. The most common...     

Age group classification and gender recognition from speech with temporal convolutional neural networks

Multimedia Tools and Applications - This paper analyses the performance of different types of Deep Neural Networks to jointly estimate age and identify gender from speech, to be applied in...     

A novel softplus linear unit for deep convolutional neural networks

Current improvements in the performance of deep neural networks are partly due to the proposition of rectified linear units. A ReLU activation function outputs zero for negative component, inducing the death of some neurons and a bias shift of the outputs, which causes oscillations and impedes learning. According to the theory that “zero mean activations improve learning ability”, a softplus linear unit (SLU) is proposed as an adaptive activation function that can speed up learning and improve performance in deep convolutional neural networks. Firstly, for the reduction of the bias shift, negative inputs are processed using the softplus function, and a general form of the SLU function is proposed. Secondly, the parameters of the positive component are fixed to control vanishing gradients. Thirdly, the rules for updating the parameters of the negative component are established to meet back- propagation requirements. Finally, we designed deep auto-encoder networks and conducted several experiments with them on the MNIST dataset for unsupervised learning. For supervised learning, we designed deep convolutional neural networks and conducted several experiments with them on the CIFAR-10 dataset. The experiments have shown faster convergence and better performance for image classification of SLU-based networks compared with rectified activation functions.     

A novel measure of target scattering randomness with application to polarimetric SAR image classification

A novel measure of target scattering randomness, called the average degree of randomness (ADoR), is introduced in this article. The proposed parameter ADoR is based on the degrees of polarization of the scattered waves using orthogonally polarized incident waves. Combining the ADoR and the Freeman decomposition, which is applied to discriminate the dominant scattering mechanism of the target, a new scheme for unsupervised classification of polarimetric synthetic aperture radar (PolSAR) images is designed. Considering that the preset intervals of the randomness measure may not fit the data distribution, an iterative classification method is developed. The effectiveness of the randomness measure and the proposed methods is demonstrated using a National Aeronautics and Space Administration (NASA)/Jet Propulsion Laboratory (JPL) AIRborne Synthetic Aperture Radar (AIRSAR) PolSAR image.     

图像分类的深度卷积神经网络模型综述

图像分类是计算机视觉中的一项重要任务,传统的图像分类方法具有一定的局限性。随着人工智能技术的发展,深度学习技术越来越成熟,利用深度卷积神经网络对图像进行分类成为研究热点,图像分类的深度卷积神经网络结构越来越多样,其性能远远好于传统的图像分类方法。本文立足于图像分类的深度卷积神经网络模型结构,根据模型发展和模型优化的历程,将深度卷积神经网络分为经典深度卷积神经网络模型、注意力机制深度卷积神经网络模型、轻量级深度卷积神经网络模型和神经网络架构搜索模型等4类,并对各类深度卷积神经网络模型结构的构造方法和特点进行了全面综述,对各类分类模型的性能进行了对比与分析。虽然深度卷积神经网络模型的结构设计越来越精妙,模型优化的方法越来越强大,图像分类准确率在不断刷新的同时,模型的参数量也在逐渐降低,训练和推理速度不断加快。然而深度卷积神经网络模型仍有一定的局限性,本文给出了存在的问题和未来可能的研究方向,即深度卷积神经网络模型主要以有监督学习方式进行图像分类,受到数据集质量和规模的限制,无监督式学习和半监督学习方式的深度卷积神经网络模型将是未来的重点研究方向之一;深度卷积神经网络模型的速度和资源消耗仍不尽人意,应用于移动式设备具有一定的挑战性;模型的优化方法以及衡量模型优劣的度量方法有待深入研究;人工设计深度卷积神经网络结构耗时耗力,神经架构搜索方法将是未来深度卷积神经网络模型设计的发展方向。     

面向SAR目标识别的深度卷积神经网络结构设计

目的 针对用于SAR （synthetic aperture radar） 目标识别的深度卷积神经网络模型结构的优化设计难题,在分析卷积核宽度对分类性能影响基础上,设计了一种适用于SAR目标识别的深度卷积神经网络结构。方法 首先基于二维随机卷积特征和具有单个隐层的神经网络模型-超限学习机分析了卷积核宽度对SAR图像目标分类性能的影响;然后,基于上述分析结果,在实现空间特征提取的卷积层中采用多个具有不同宽度的卷积核提取目标的多尺度局部特征,设计了一种适用于SAR图像目标识别的深度模型结构;最后,在对MSTAR （moving and stationary target acquisition and recognition） 数据集中的训练样本进行样本扩充基础上,设定了深度模型训练的超参数,进行了深度模型参数训练与分类性能验证。结果 实验结果表明,对于具有较强相干斑噪声的SAR图像而言,采用宽度更大的卷积核能够提取目标的局部特征,提出的模型因能从输入图像提取目标的多尺度局部特征,对于10类目标的分类结果（包含非变形目标和变形目标两种情况）接近或优于已知文献的最优分类结果,目标总体分类精度分别达到了98.39%和97.69%,验证了提出模型结构的有效性。结论 对于SAR图像目标识别,由于与可见光图像具有不同的成像机理,应采用更大的卷积核来提取目标的空间特征用于分类,通过对深度模型进行优化设计能够提高SAR图像目标识别的精度。     

A multi-class skin Cancer classification using deep convolutional neural networks

Skin Cancer accounts for one-third of all diagnosed cancers worldwide. The prevalence of skin cancers have been rising over the past decades. In recent years, use of dermoscopy has enhanced the diagnostic capability of skin cancer. The accurate diagnosis of skin cancer is challenging for dermatologists as multiple skin cancer types may appear similar in appearance. The dermatologists have an average accuracy of 62% to 80% in skin cancer diagnosis. The research community has been made significant progress in developing automated tools to assist dermatologists in decision making. In this work, we propose an automated computer-aided diagnosis system for multi-class skin (MCS) cancer classification with an exceptionally high accuracy. The proposed method outperformed both expert dermatologists and contemporary deep learning methods for MCS cancer classification. We performed fine-tuning over seven classes of HAM10000 dataset and conducted a comparative study to analyse the performance of five pre-trained convolutional neural networks (CNNs) and four ensemble models. The maximum accuracy of 93.20% for individual model amongst the set of models whereas maximum accuracy of 92.83% for ensemble model is reported in this paper. We propose use of ResNeXt101 for the MCS cancer classification owing to its optimized architecture and ability to gain higher accuracy.

     

Multi‐representational convolutional neural networks for text classification

Various studies have demonstrated that convolutional neural networks (CNNs) can be directly applied to different levels of text embedding, such as character‐, word‐, or document‐levels. However, the effectiveness of different embeddings is limited in the reported result and there is a lack of clear guidance on some aspects of their use, including choosing the proper level of embedding and switching word semantics from one domain to another when appropriate. In this paper, we propose a new architecture of CNN based on multiple representations for text classification, by constructing multiple planes so that more information can be dumped into the networks, such as different parts of text obtained through named entity recognizer or part‐of‐speech tagging tools, different levels of text embedding, or contextual sentences. Various large‐scale, domain‐specific datasets are used to validate the proposed architecture. Tasks analyzed include ontology document classification, biomedical event categorization, and sentiment analysis, showing that multi‐representational CNNs, which learns to focus attention to specific representations of text, can obtain further gains in performance over state‐of‐the‐art deep neural network models.     

Smartphone-based bulky waste classification using convolutional neural networks

Multimedia Tools and Applications - The rapid urbanization process is escalating the urban waste problem, and ineffective management has worsened the issue, leading to severe consequences to the...     

A novel companion objective function for regularization of deep convolutional neural networks

Regularization is an essential technique discussed in an attempt to solve the overfitting problem in deep convolutional neural networks (CNNs). In this paper, we proposed a novel companion objective function as a regularization strategy to boost the classification performance in deep CNNs. Three aspects of this companion objective function are studied. Firstly, we proposed two kinds of auxiliary supervision for convolutional filters and non-linear activations respectively in the companion objective function. Both of them enhanced the performance by aleviating the overfitting problem and auxiliary supervision for non-linear activations provided more efficiency. Secondly, regularization of auxiliary supervision in the pre-training phrase is discussed. With the assistance of auxiliary supervision, CNNs could obtain a more favorable initialization for end-to-end supervised fine-tuning. Finally, this companion objective function is verified to be compatible with other regularization strategies such as dropout and data augmentation. Experimental results on benchmark datasets (CIFAR-10 and CIFAR-100) demonstrated advantages of our proposed companion objective function as a regularization approach.     

SAR变体目标识别的卷积神经网络法

目的 深度学习已经大量应用于合成孔径宽达（SAR）图像目标识别领域,但大多数工作是基于MSTAR数据集的标准操作条件展开研究。当将深度学习应用于同类含变体目标时,例如T72子类,由于目标间差异小,所以仍存在着较大的挑战。本文从极大限度地保留SAR图像输入特征出发,设计一种适用于SAR变体目标识别的深度卷积神经网络结构。方法 设计网络主要由多尺度空间特征提取模块和DenseNet中的稠密块、转移层构成。多尺度特征提取模块置于网络底层,通过使用尺寸分别为1×1、3×3、5×5、7×7、9×9的卷积核,提取丰富空间特征的同时保留输入图像信息。为使输入图像信息更加有效地向后传递,基于DenseNet中的稠密块和转移层进行后续网络层设计。在对训练样本进行样本扩充基础上,分析了输入图像分辨率及目标存在平移和不同噪声水平等情况对模型识别精度的影响,与用于SAR图像目标识别的深度模型识别精度在标准操作条件下进行了对比分析。结果 实验结果表明,对T72 8类变体目标进行分类,设计的模型能够取得95.48%的识别精度,在存在目标平移和不同噪声水平情况下,平均识别精度分别达到了94.61%和86.36%。对10类目标（包括不含变体和含变体情况）在进行数据增强的情况下进行模型训练与测试,分别达到了99.38%和98.81%的识别精度,略优于其他对比模型结构识别精度。结论 提出的模型可以充分利用输入图像以及各卷积层输出的特征,学习目标图像的细节差异,不仅适用于SAR图像变体目标的识别任务,同时在标准操作条件下的识别任务也取得了较高的识别结果。     

Using convolutional neural networks for classification of malware represented as images

Journal of Computer Virology and Hacking Techniques - The number of malicious files detected every year are counted by millions. One of the main reasons for these high volumes of different files is...     

Improvise approach for respiratory pathologies classification with multilayer convolutional neural networks

Multimedia Tools and Applications - Every respiratory-related checkup includes audio samples collected from the individual, collected through different tools (sonograph, stethoscope). This audio is...     

A transfer learning-based novel fusion convolutional neural network for breast cancer histology classification

Multimedia Tools and Applications - To train a convolutional neural network (CNN) from scratch is not suitable for medical image tasks with insufficient data. Benefiting from the transfer learning,...     

Exploiting multi-channels deep convolutional neural networks for multivariate time series classification

Time series classification is related to many different domains, such as health informatics, finance, and bioinformatics. Due to its broad applications, researchers have developed many algorithms for this kind of tasks, e.g., multivariate time series classification. Among the classification algorithms, k-nearest neighbor (k-NN) classification (particularly 1-NN) combined with dynamic time warping (DTW) achieves the state of the art performance. The deficiency is that when the data set grows large, the time consumption of 1-NN with DTWwill be very expensive. In contrast to 1-NN with DTW, it is more efficient but less effective for feature-based classification methods since their performance usually depends on the quality of hand-crafted features. In this paper, we aim to improve the performance of traditional feature-based approaches through the feature learning techniques. Specifically, we propose a novel deep learning framework, multi-channels deep convolutional neural networks (MC-DCNN), for multivariate time series classification. This model first learns features from individual univariate time series in each channel, and combines information from all channels as feature representation at the final layer. Then, the learnt features are applied into a multilayer perceptron (MLP) for classification. Finally, the extensive experiments on real-world data sets show that our model is not only more efficient than the state of the art but also competitive in accuracy. This study implies that feature learning is worth to be investigated for the problem of time series classification.