Few-Shot Learning with Generative Adversarial Networks Based on WOA13 Data

In recent years, extreme weather events accompanying the global warming have occurred frequently, which brought significant impact on national economic and social development. The ocean is an important member of the climate system and plays an important role in the occurrence of climate anomalies. With continuous improvement of sensor technology, we use sensors to acquire the ocean data for the study on resource detection and disaster prevention, etc. However, the data acquired by the sensor is not enough to be used directly by researchers, so we use the Generative Adversarial Network (GAN) to enhance the ocean data. We use GAN to process WOA13 dataset and use ResNet to determine if there is a thermocline layer in a sea area. We compare the classification results of the enhanced datasets of different orders of magnitude with the classification results of the original datasets. The experimental result shows that the dataset processed by GAN has a higher accuracy. GAN has a certain enhancement effect to marine data. Gan increased the accuracy of the WOA dataset from 0.91 to 0.93. At the same time, the experimental results also show that too much data cannot continue to enhance the accuracy of WOA in ResNet.     

Data Augmentation and Random Multi-Model Deep Learning for Data Classification

In the machine learning (ML) paradigm, data augmentation serves as a regularization approach for creating ML models. The increase in the diversification of training samples increases the generalization capabilities, which enhances the prediction performance of classifiers when tested on unseen examples. Deep learning (DL) models have a lot of parameters, and they frequently overfit. Effectively, to avoid overfitting, data plays a major role to augment the latest improvements in DL. Nevertheless, reliable data collection is a major limiting factor. Frequently, this problem is undertaken by combining augmentation of data, transfer learning, dropout, and methods of normalization in batches. In this paper, we introduce the application of data augmentation in the field of image classification using Random Multi-model Deep Learning (RMDL) which uses the association approaches of multiDL to yield random models for classification. We present a methodology for using Generative Adversarial Networks (GANs) to generate images for data augmenting. Through experiments, we discover that samples generated by GANs when fed into RMDL improve both accuracy and model efficiency. Experimenting across both MNIST and CIAFAR-10 datasets show that, error rate with proposed approach has been decreased with different random models.     

注意力机制对生成对抗网络语音增强迁移学习模型的影响

基于深度学习的语音增强模型对训练集外语言语音和噪声进行降噪时,性能明显下降.为了解决这一问题,提出一种引入注意力机制的生成对抗网络(Generative Adversarial Network,GAN)语音增强迁移学习模型.在生成对抗语音增强模型的判别模型中引入注意力机制,以高资源场景下的大量语音数据训练得到的语音增强...     

Improving Brain Tumor Classification with Deep Learning UsingSyntheticData

Deep learning (DL) techniques, which do not need complex pre-processing and feature analysis, are used in many areas of medicine and achieve promising results. On the other hand, in medical studies, a limited dataset decreases the abstraction ability of the DL model. In this context, we aimed to produce synthetic brain images including three tumor types (glioma, meningioma, and pituitary), unlike traditional data augmentation methods, and classify them with DL. This study proposes a tumor classification model consisting of a Dense Convolutional Network (DenseNet121)-based DL model to prevent forgetting problems in deep networks and delay information flow between layers. By comparing models trained on two different datasets, we demonstrated the effect of synthetic images generated by Cycle Generative Adversarial Network (CycleGAN) on the generalization of DL. One model is trained only on the original dataset, while the other is trained on the combined dataset of synthetic and original images. Synthetic data generated by CycleGAN improved the best accuracy values for glioma, meningioma, and pituitary tumor classes from 0.9633, 0.9569, and 0.9904 to 0.9968, 0.9920, and 0.9952, respectively. The developed model using synthetic data obtained a higher accuracy value than the related studies in the literature. Additionally, except for pixel-level and affine transform data augmentation, synthetic data has been generated in the figshare brain dataset for the first time.     

A Scalable Method of Maintaining Order Statistics for Big Data Stream

Recently, there are some online quantile algorithms that work on how to analyze the order statistics about the high-volume and high-velocity data stream, but the drawback of these algorithms is not scalable because they take the GK algorithm as the subroutine, which is not known to be mergeable. Another drawback is that they can’t maintain the correctness, which means the error will increase during the process of the window sliding. In this paper, we use a novel data structure to store the sketch that maintains the order statistics over sliding windows. Therefore three algorithms have been proposed based on the data structure. And the fixed-size window algorithm can keep the sketch of the last W elements. It is also scalable because of the mergeable property. The time-based window algorithm can always keep the sketch of the data in the last T time units. Finally, we provide the window aggregation algorithm which can help extend our algorithm into the distributed system. This provides a speed performance boost and makes it more suitable for modern applications such as system/network monitoring and anomaly detection. The experimental results show that our algorithm can not only achieve acceptable performance but also can actually maintain the correctness and be mergeable.     

多网络联合的红外与可见光图像融合算法研究

目的针对红外与可见光图像在融合过程中,融合图像失真以及可见光图像信息融合不足的问题,提出一种联合多网络结构的红外与可见光图像融合算法。方法首先采用基于密集残差连接的编码器对输入的红外与可见光图像进行特征提取,然后利用融合策略对得到的特征图进行融合,最后将融合后的特征图送入基于GAN网络的解码器中。结果通过与可见光图像对抗优化训练,使得融合后的图像保留了更多可见光图像的细节、背景信息,增强了图像的视觉效果。结论实验表明,与现有的融合算法相比,该算法达到了更好的实验效果,在主观感知和客观评价上都具有更好的表现力。     

Pseudo NLP Joint Spam Classification Technique for Big Data Cluster

Spam mail classification considered complex and error-prone task in the distributed computing environment. There are various available spam mail classification approaches such as the naive Bayesian classifier, logistic regression and support vector machine and decision tree, recursive neural network, and long short-term memory algorithms. However, they do not consider the document when analyzing spam mail content. These approaches use the bag-of-words method, which analyzes a large amount of text data and classifies features with the help of term frequency-inverse document frequency. Because there are many words in a document, these approaches consume a massive amount of resources and become infeasible when performing classification on multiple associated mail documents together. Thus, spam mail is not classified fully, and these approaches remain with loopholes. Thus, we propose a term frequency topic inverse document frequency model that considers the meaning of text data in a larger semantic unit by applying weights based on the document’s topic. Moreover, the proposed approach reduces the scarcity problem through a frequency topic-inverse document frequency in singular value decomposition model. Our proposed approach also reduces the dimensionality, which ultimately increases the strength of document classification. Experimental evaluations show that the proposed approach classifies spam mail documents with higher accuracy using individual document-independent processing computation. Comparative evaluations show that the proposed approach performs better than the logistic regression model in the distributed computing environment, with higher document word frequencies of 97.05%, 99.17% and 96.59%.     

High Security for De-Duplicated Big Data Using Optimal SIMON Cipher

Cloud computing offers internet location-based affordable, scalable, and independent services. Cloud computing is a promising and a cost-effective approach that supports big data analytics and advanced applications in the event of forced business continuity events, for instance, pandemic situations. To handle massive information, clusters of servers are required to assist the equipment which enables streamlining the widespread quantity of data, with elevated velocity and modified configurations. Data deduplication model enables cloud users to efficiently manage their cloud storage space by getting rid of redundant data stored in the server. Data deduplication also saves network bandwidth. In this paper, a new cloud-based big data security technique utilizing dual encryption is proposed. The clustering model is utilized to analyze the Deduplication process hash function. Multi kernel Fuzzy C means (MKFCM) was used which helps cluster the data stored in cloud, on the basis of confidence data encryption procedure. The confidence finest data is implemented in homomorphic encryption data wherein the Optimal SIMON Cipher (OSC) technique is used. This security process involving dual encryption with the optimization model develops the productivity mechanism. In this paper, the excellence of the technique was confirmed by comparing the proposed technique with other encryption and clustering techniques. The results proved that the proposed technique achieved maximum accuracy and minimum encryption time.     

HDLIDP: A Hybrid Deep Learning Intrusion Detection and Prevention Framework

Distributed denial-of-service (DDoS) attacks are designed to interrupt network services such as email servers and webpages in traditional computer networks. Furthermore, the enormous number of connected devices makes it difficult to operate such a network effectively. Software defined networks (SDN) are networks that are managed through a centralized control system, according to researchers. This controller is the brain of any SDN, composing the forwarding table of all data plane network switches. Despite the advantages of SDN controllers, DDoS attacks are easier to perpetrate than on traditional networks. Because the controller is a single point of failure, if it fails, the entire network will fail. This paper offers a Hybrid Deep Learning Intrusion Detection and Prevention (HDLIDP) framework, which blends signature-based and deep learning neural networks to detect and prevent intrusions. This framework improves detection accuracy while addressing all of the aforementioned problems. To validate the framework, experiments are done on both traditional and SDN datasets; the findings demonstrate a significant improvement in classification accuracy.     

基于大数据挖掘的老年手杖个性化定制研究

目的 随着互联网的发展,用户评论快速增长,利用这一海量数据进行文本分析,结合Kano模型,以此来获取更加全面的用户定制需求。方法 提出了一种基于大数据评论文本挖掘的方法,来获取老年手杖个性化定制需求。首先将老年手杖分为三大不同产品等级,挑选典型产品,爬取用户评论;其次通过文本的对应分析获取不同等级手杖用户需求的差异;接着利用LDA模型、结合德尔菲专家法获取用户需求族群;最后利用Kano模型进行需求等级划分,并结合Fisher精确检验进行差异显著度检验。结果 识别出老年手杖的基本型、期望型、兴奋型不同等级用户需求,以指导老年手杖个性化定制界面的设计。结论 结果表明大数据挖掘与Kano模型相结合的方法,能够有效地获取用户个性化需求层级,并指导定制平台的搭建,为产品个性化定制平台的设计提供科学依据。     

Deep Learning Based Optimal Multimodal Fusion Framework for Intrusion Detection Systems for Healthcare Data

Data fusion is a multidisciplinary research area that involves different domains. It is used to attain minimum detection error probability and maximum reliability with the help of data retrieved from multiple healthcare sources. The generation of huge quantity of data from medical devices resulted in the formation of big data during which data fusion techniques become essential. Securing medical data is a crucial issue of exponentially-pacing computing world and can be achieved by Intrusion Detection Systems (IDS). In this regard, since singular-modality is not adequate to attain high detection rate, there is a need exists to merge diverse techniques using decision-based multimodal fusion process. In this view, this research article presents a new multimodal fusion-based IDS to secure the healthcare data using Spark. The proposed model involves decision-based fusion model which has different processes such as initialization, pre-processing, Feature Selection (FS) and multimodal classification for effective detection of intrusions. In FS process, a chaotic Butterfly Optimization (BO) algorithm called CBOA is introduced. Though the classic BO algorithm offers effective exploration, it fails in achieving faster convergence. In order to overcome this, i.e., to improve the convergence rate, this research work modifies the required parameters of BO algorithm using chaos theory. Finally, to detect intrusions, multimodal classifier is applied by incorporating three Deep Learning (DL)-based classification models. Besides, the concepts like Hadoop MapReduce and Spark were also utilized in this study to achieve faster computation of big data in parallel computation platform. To validate the outcome of the presented model, a series of experimentations was performed using the benchmark NSLKDDCup99 Dataset repository. The proposed model demonstrated its effective results on the applied dataset by offering the maximum accuracy of 99.21%, precision of 98.93% and detection rate of 99.59%. The results assured the betterment of the proposed model.     

Data Analytics and Machine Learning for Smart Process Manufacturing: Recent Advances and Perspectives in the Big Data Era

Safe, efficient, and sustainable operations and control are primary objectives in industrial manufacturing processes. State-of-the-art technologies heavily rely on human intervention, thereby showing apparent limitations in practice. The burgeoning era of big data is influencing the process industries tremendously, providing unprecedented opportunities to achieve smart manufacturing. This kind of manufacturing requires machines to not only be capable of relieving humans from intensive physical work, but also be effective in taking on intellectual labor and even producing innovations on their own. To attain this goal, data analytics and machine learning are indispensable. In this paper, we review recent advances in data analytics and machine learning applied to the monitoring, control, and optimization of industrial processes, paying particular attention to the interpretability and functionality of machine learning models. By analyzing the gap between practical requirements and the current research status, promising future research directions are identified.     

A Distributed Approach of Big Data Mining for Financial Fraud Detection in a Supply Chain

Supply Chain Finance (SCF) is important for improving the effectiveness of supply chain capital operations and reducing the overall management cost of a supply chain. In recent years, with the deep integration of supply chain and Internet, Big Data, Artificial Intelligence, Internet of Things, Blockchain, etc., the efficiency of supply chain financial services can be greatly promoted through building more customized risk pricing models and conducting more rigorous investment decision-making processes. However, with the rapid development of new technologies, the SCF data has been massively increased and new financial fraud behaviors or patterns are becoming more covertly scattered among normal ones. The lack of enough capability to handle the big data volumes and mitigate the financial frauds may lead to huge losses in supply chains. In this article, a distributed approach of big data mining is proposed for financial fraud detection in a supply chain, which implements the distributed deep learning model of Convolutional Neural Network (CNN) on big data infrastructure of Apache Spark and Hadoop to speed up the processing of the large dataset in parallel and reduce the processing time significantly. By training and testing on the continually updated SCF dataset, the approach can intelligently and automatically classify the massive data samples and discover the fraudulent financing behaviors, so as to enhance the financial fraud detection with high precision and recall rates, and reduce the losses of frauds in a supply chain.     

大数据下的深度学习研究

给出了大数据和机器学习的子领域——深度学习的概念,阐述了深度学习对获取大数据中的有价值信息的重要作用。描述了大数据下利用图像处理单元(GPU)进行并行运算的深度学习框架,对其中的大规模卷积神经网络(CNN)、大规模深度置信网络(DBN)和大规模递归神经网络(RNN)进行了重点论述。分析了大数据的容量、多样性、速率特征,介绍了大规模数据、多样性数据、高速率数据下的深度学习方法。展望了大数据背景下深度学习的发展前景,指出在不远的将来,大数据与深度学习融合的技术将会在计算机视觉、机器智能等多个领域获得突破性进展。     

基于CNN-集成学习的多风电机组故障诊断

海上风电场地处偏远环境,长期受到盐碱腐蚀。为解决风电机组运行过程中产生的多种故障检测识别问题,在传统卷积神经网络LeNet-5的基础上构建模型。该模型采用ReLU函数作为激活函数,增加了卷积层、池化层和全连接层。针对风电机组的监督控制和数据采集(supervisory control and data acquisition,SCADA)系统及状态监控(condition monitoring,CM)系统所提供的数据集,进行多元类别故障诊断。并对多台风电机组进行聚类分析,应用集成学习方法,构建多风电机组故障诊断模型。实验表明,所提方法取得了97%~99%的诊断精度。通过将实验结果与其他算法进行对比,验证了该方法的有效性。     

面向智能交互产品的创意服务设计模式研究与构建

目的 探索数智时代下创意服务设计产业应用于智能交互产品创新的模式机制与路径。方法 阐述创意服务产业时代的内涵与特征,梳理数字时代智能交互产品的创意服务需求,并从产品的品质创新需求、创新绩效提升需求和经济价值转化路径优化需求三个视角剖析创意服务设计模式的创新策略。结果 提出“多元设计大数据创新-智能生成式设计工具优化-利益相关者协同设计平台重构”三段式创意服务设计模式,并结合案例进行分析。结论 通过对创意服务设计的数据资源创新、设计工具的智能优化和协同创新的价值链重构,可有效提升智能交互产品设计结果的输出品质和数量,优化设计迭代周期,实现设计驱动的产品价值创新绩效和经济价值效益的双向提升。通过对知识生产力要素进行数智化整合重构,构建全链设计要素集成的智能辅助设计工具体系,提升设计师群体的创新能力,从而为社会经济价值提升带来强劲的可持续创新设计新动力。