Deep learning (DL) techniques have the potential of making communication systems more efficient and solving many problems in the physical layer. In this paper, an optical wireless communications (OWC) system based on ...
详细信息
ISBN:
(纸本)9781538661543
Deep learning (DL) techniques have the potential of making communication systems more efficient and solving many problems in the physical layer. In this paper, an optical wireless communications (OWC) system based on visible light communications (VLC) technology is implemented using an autoencoder (AE). The proposed system is tested in different scenarios using various AE parameters and applied on an indoor VLC model. Bit error rate (BER) is evaluated with respect to the signal-to-noise-ratio (SNR) values at different locations within the room. To validate the proposed system, theoretical results are compared to the simulated values. The bit-error performance demonstrates the viability of DL techniques in VLC systems.
With the growing usage of credit card transactions, financial fraud crimes have also been drastically increased leading to the loss of huge amounts in the finance industry. Having an efficient fraud detection method h...
详细信息
ISBN:
(纸本)9781538670453
With the growing usage of credit card transactions, financial fraud crimes have also been drastically increased leading to the loss of huge amounts in the finance industry. Having an efficient fraud detection method has become a necessity for all banks in order to minimize such losses. In fact, credit card fraud detection system involves a major challenge: the credit card fraud data sets are highly imbalanced since the number of fraudulent transactions is much smaller than the legitimate ones. Thus, many of traditional classifiers often fail to detect minority class objects for these skewed data sets. This paper aims first: to enhance classified performance of the minority of credit card fraud instances in the imbalanced data set, for that we propose a sampling method based on the K-means clustering and the genetic algorithm. We used K-means algorithm to cluster and group the minority kind of sample, and in each cluster we use the genetic algorithm to gain the new samples and construct an accurate fraud detection classifier.
Aiming at the difficulty of semantic gap in content-based image search (CBIR), inspired by the convolutional neural network (CNN) in image classification and detection, this paper proposes a simple and effective hybri...
详细信息
ISBN:
(纸本)9781538674161
Aiming at the difficulty of semantic gap in content-based image search (CBIR), inspired by the convolutional neural network (CNN) in image classification and detection, this paper proposes a simple and effective hybrid model of deep convolutional network and autoencoder network. This model uses the CNN network to extract the high-level semantic features of the image, then uses the depth autoencoder network to reduce the dimension of the extracted image features, and compresses the features into a 128-bit vector representation. Nearest Neighbor Search (ANN) is an effective strategy for large-scale image retrieval. This paper uses the annoy algorithm to calculate the similarity between the query image and the index tree, and outputs them in descending order of similarity. Experimental results show that the proposed method outperforms some of the latest deep-network image retrieval algorithms on the CIFAR-10 and MNIST datasets. In the TOP10 image search, the MNSIT dataset can obtain 100% accuracy. In the CIFAR dataset experiment, the accuracy and recall rate of the CIFAR4 dataset are as high as 99.9%, and the accuracy and recall rate of the CIF'AR10 dataset reach respectively 97.2% and 98.1% In addition, the size of the convolutional network's parameters and the size of the index are optimized compared to the previous model, so that the effect of second-level real-time response can be achieved in the 10,000-level image search.
Radar signals are time series that have pulse repetition interval, pulse width and pulse amplitude as their features. After reception of them by electronic warfare systems, their features are classified and kept in a ...
详细信息
ISBN:
(纸本)9781538615010
Radar signals are time series that have pulse repetition interval, pulse width and pulse amplitude as their features. After reception of them by electronic warfare systems, their features are classified and kept in a database. This procedure brings vision for the system user if the same signal is received again in the future. For this classification purpose, three algorithms were implemented. The first one is a combined network consisting of a Convolutional Neural Network (CNN) and a Long-Short Time Memory (LSTM), the second is a Hybrid Network including the first network and a CNN which enables parallel training having histogram data as its input. The last algorithm is Stacked autoencoder. Performance analysis was made on real radar data. The best performer is Hybrid Network which reached 99.6% accuracy as it involves histogram data usage and an LSTM for the time series problem. Convolutional LSTM reached 98.3% while Stacked autoencoder had 87% accuracy.
In this paper,we describe a intrusion detection algorithm based on deep learning for industrial control networks,aiming at the security problem of industrial control *** learning is a kind of intelligent algorithm and...
详细信息
In this paper,we describe a intrusion detection algorithm based on deep learning for industrial control networks,aiming at the security problem of industrial control *** learning is a kind of intelligent algorithm and has the ability of automatically *** use self-learning to enhance the experience and dynamic classification *** ideology of deep learning is similar to the idea of intrusion detection to improve the detection rate and reduce the rate of false through learning,a sparse auto-encoder-extreme learning machine intrusion detection model is proposed for the problem of intrusion detection *** uses deep learning autoencoder to combine the coefficient penalty and reconstruction loss of the encode layer to extract the features of high-dimensional data during the training model,and then uses the extreme learning machine to quickly and effectively classify the extracted *** accuracy of the algorithm is verified by the industrial control intrusion detection standard data *** experimental results verify that the method can effectively improve the performance of the intrusion detection system and reduce the false alarm rate.
Interferometric Synthetic Aperture Radar (InSAR) imagery for estimating ground movement, based on microwaves reflected off ground targets is gaining increasing importance in remote sensing. However, noise corrupts mic...
详细信息
ISBN:
(纸本)9781538647073
Interferometric Synthetic Aperture Radar (InSAR) imagery for estimating ground movement, based on microwaves reflected off ground targets is gaining increasing importance in remote sensing. However, noise corrupts microwave reflections received at satellite and contaminates the signal's wrapped phase. We introduce Convolutional Neural Networks (CNNs) to this problem domain and show the effectiveness of autoencoder CNN architectures to learn InSAR image denoising filters in the absence of clean ground truth images, and for artefact reduction in estimated coherence through intelligent preprocessing of training data. We compare our results with four established methods to illustrate superiority of proposed method.
Left ventricular (LV) volumes, and emptying and filling function remain important indices in conditions such as heart failure. These parameters are derived from the volume curve contained by the inner border of the LV...
详细信息
ISBN:
(纸本)9781538675687
Left ventricular (LV) volumes, and emptying and filling function remain important indices in conditions such as heart failure. These parameters are derived from the volume curve contained by the inner border of the LV of the heart, throughout the emptying and filling phases of the cardiac cycle, and the peak emptying and filling rates. The gold standard uses the Simpson rule to estimate volume from stacks of short axis images acquired using cine MRI. In this study, a deep learning, automated supervised approach to estimate ventricular volumes is introduced. Unlike prior methods that required hand-crafted image features to segment the inner contour, the proposed approach uses an automatically selected region of interest (ROI), and intelligently determines the optimum features directly from the ROI information. These derived features are then inputted into a deep learning regression model, with the estimated volume as the output results.
Video captioning is a challenging task owing to the complexity of understanding the copious visual information in videos and describing it using natural language. Different from previous work that encodes video inform...
详细信息
ISBN:
(纸本)9781450356657
Video captioning is a challenging task owing to the complexity of understanding the copious visual information in videos and describing it using natural language. Different from previous work that encodes video information using a single flow, in this work, we introduce a novel Sibling Convolutional Encoder (SibNet) for video captioning, which utilizes a two-branch architecture to collaboratively encode videos. The first content branch encodes the visual content information of the video via autoencoder, and the second semantic branch encodes the semantic information by visual-semantic joint embedding. Then both branches are effectively combined with soft-attention mechanism and finally fed into a RNN decoder to generate captions. With our SibNet explicitly capturing both content and semantic information, the proposed method can better represent the rich information in videos. Extensive experiments on YouTube2Text and MSR-VTT datasets validate that the proposed architecture outperforms existing methods by a large margin across different evaluation metrics.
Network Intrusion Detection Systems (NIDSs) are increasingly crucial due to the expansion of computer networks. Detection techniques based on machine learning have attracted extensive attention for their capability to...
详细信息
ISBN:
(纸本)9783030000127;9783030000110
Network Intrusion Detection Systems (NIDSs) are increasingly crucial due to the expansion of computer networks. Detection techniques based on machine learning have attracted extensive attention for their capability to detect novel attacks. However, they require a large amount of labeled training data to train an effective model, which is difficult and expensive to obtain. To this effect, it is critically important to build models which can learn from unlabeled or partially-labeled data. In this paper, we propose an autoencoder-based framework, i. e., SU-IDS, for semi-supervised and unsupervised network intrusion detection. The framework augments the usual clustering (or classification) loss with an auxiliary loss of autoencoder, and thus achieves a better performance. The experimental results on the classic NSL-KDD dataset and the modern CICIDS2017 dataset show the superiority of our proposed models.
As an emerging technology, device-free localization (DFL), using radio frequency (RF) sensor networks to detect targets who do not carry any attached devices, has spawned extensive applications. Many existing works fo...
详细信息
ISBN:
(纸本)9781538666500
As an emerging technology, device-free localization (DFL), using radio frequency (RF) sensor networks to detect targets who do not carry any attached devices, has spawned extensive applications. Many existing works formulate DFL as a classification problem, and a key problem is how to extract discriminative features to characterize the raw wireless signal. In this paper, we present an autoencoder-based deep neural network for feature extraction, moreover, multiple Gaussian Bernoulli restricted Boltzmann machines (GBRBMs) are utilized for pre-training and dimension reduction. Experiment results show that this method of GBRBM-based autoencoder (GBRBM-AE) can achieve a high accuracy and efficient performance, which outperforms the conventional autoencoder. When the dimensions of input data are reduced from 784 to 20 dims, our algorithm can maintain a high accuracy of 97.1% and is robust to noise with SNR = 5dB.
暂无评论