Single-channel Network Research Articles

Dual-branch network with fused Mel features for logic-manipulated speech detection

In recent years, logic-manipulated speech detection has predominantly relied on single-feature and single-channel networks, which fail to consider a broader range of acoustic features. To address this limitation, we propose a novel approach based on a dual-branch network that incorporates fused Mel features for detecting logic-manipulated speech. The structure of our network consists of three main components. Firstly, we extract Mel spectrograms and Mel-frequency cepstral coefficients (MFCC) to address the deficiencies caused by the reduction in the information dimension. Then, to better utilize these two types of features, we construct a dual-branch network framework that enables the separate learning of temporal and frequency domain characteristics. Finally, at the backend of the network, we fuse the feature vectors for decision-making. Experimental evaluations conducted on the ASVspoof2019LA dataset demonstrate the effectiveness and superiority of our proposed method compared to state-of-the-art models and other acoustic features.

Multi-channel high-order network representation learning research.

The existing network representation learning algorithms mainly model the relationship between network nodes based on the structural features of the network, or use text features, hierarchical features and other external attributes to realize the network joint representation learning. Capturing global features of the network allows the obtained node vectors to retain more comprehensive feature information during training, thereby enhancing the quality of embeddings. In order to preserve the global structural features of the network in the training results, we employed a multi-channel learning approach to perform high-order feature modeling on the network. We proposed a novel algorithm for multi-channel high-order network representation learning, referred to as the Multi-Channel High-Order Network Representation (MHNR) algorithm. This algorithm initially constructs high-order network features from the original network structure, thereby transforming the single-channel network representation learning process into a multi-channel high-order network representation learning process. Then, for each single-channel network representation learning process, the novel graph assimilation mechanism is introduced in the algorithm, so as to realize the high-order network structure modeling mechanism in the single-channel network representation learning. Finally, the algorithm integrates the multi-channel and single-channel mechanism of high-order network structure joint modeling, realizing the efficient use of network structure features and sufficient modeling. Experimental results show that the node classification performance of the proposed MHNR algorithm reaches a good order on Citeseer, Cora, and DBLP data, and its node classification performance is better than that of the comparison algorithm used in this paper. In addition, when the vector length is optimized, the average classification accuracy of nodes of the proposed algorithm is up to 12.24% higher than that of the DeepWalk algorithm. Therefore, the node classification performance of the proposed algorithm can reach the current optimal order only based on the structural features of the network under the condition of no external feature supplementary modeling.

Wildfire Detection via a Dual-Channel CNN with Multi-Level Feature Fusion

Forest fires have devastating impacts on ecology, the economy, and human life. Therefore, the timely detection and extinguishing of fires are crucial to minimizing the losses caused by these disasters. A novel dual-channel CNN for forest fires is proposed in this paper based on multiple feature enhancement techniques. First, the features’ semantic information and richness are enhanced by repeatedly fusing deep and shallow features extracted from the basic network model and integrating the results of multiple types of pooling layers. Second, an attention mechanism, the convolutional block attention module, is used to focus on the key details of the fused features, making the network more efficient. Finally, two improved single-channel networks are merged to obtain a better-performing dual-channel network. In addition, transfer learning is used to address overfitting and reduce time costs. The experimental results show that the accuracy of the proposed model for fire recognition is 98.90%, with a better performance. The findings from this study can be applied to the early detection of forest fires, assisting forest ecosystem managers in developing timely and scientifically informed defense strategies to minimize the damage caused by fires.

A graph neural network-enhanced knowledge graph framework for intelligent analysis of policing cases.

In this paper, we model a knowledge graph based on graph neural networks, conduct an in-depth study on building knowledge graph embeddings for policing cases, and design a graph neural network-enhanced knowledge graph framework. In detail, we use the label propagation algorithm (LPA) to assist the convolutional graph network (GCN) in training the edge weights of the knowledge graph to construct a policing case prediction method. This improves the traditional convolutional neural network from a single-channel network to a multichannel network to accommodate the multiple feature factors of policing cases. In addition, this expands the perceptual field of the convolutional neural network to improve prediction accuracy. The experimental results show that the multichannel convolutional neural network's prediction accuracy can reach 87.7%. To ensure the efficiency of the security case analysis network, an efficient pairwise feature extraction base module is added to enhance the backbone network, which reduces the number of parameters of the whole network and decreases the complexity of operations. We experimentally demonstrate that this method achieves a better balance of efficiency and performance by obtaining approximate results with 53.5% fewer floating-point operations and 70.2% fewer number parameters than its contemporary work.

A vegetation classification method based on improved dual-way branch feature fusion U-net.

Aiming at the problems of complex structure parameters and low feature extraction ability of U-Net used in vegetation classification, a deep network with improved U-Net and dual-way branch input is proposed. Firstly, The principal component analysis (PCA) is used to reduce the dimension of hyperspectral remote sensing images, and the effective bands are obtained. Secondly, the depthwise separable convolution and residual connections are combined to replace the common convolution layers of U-Net for depth feature extraction to ensure classification accuracy and reduce the complexity of network parameters. Finally, normalized difference vegetation index (NDVI), gray level co-occurrence matrix (GLCM) and edge features of hyperspectral remote sensing images are extracted respectively. The above three artificial features are fused as one input, and PCA dimension reduction features are used as another input. Based on the improved U-net, a dual-way vegetation classification model is generated. Taking the hyperspectral remote sensing image of Matiwan Village, Xiong'an, Beijing as the experimental object, the experimental results show that the precision and recall of the improved U-Net are significantly improved with the residual structure and depthwise separable convolution, reaching 97.13% and 92.36% respectively. In addition, in order to verify the effectiveness of artificial features and dual-way branch design, the accuracy of single channel and the dual-way branch are compared. The experimental results show that artificial features in single channel network interfere with the original hyperspectral data, resulting in reduction of the recognition accuracy. However, the accuracy of the dual-way branch network has been improved, reaching 98.67%. It shows that artificial features are effective complements of network features.

Dual-channel asymmetric convolutional neural network for an efficient retinal blood vessel segmentation in eye fundus images

The morphological properties of retinal vessels are closely related to the diagnosis of ophthalmic diseases. However, many problems in retinal images, such as complicated directions of vessels and difficult recognition of capillaries, bring challenges to the accurate segmentation of retinal blood vessels. Thus, we propose a new retinal blood vessel segmentation method based on a dual-channel asymmetric convolutional neural network (CNN). First, we construct the thick and thin vessel extraction module based on the morphological differences in retinal vessels. A two-dimensional (2D) Gabor filter is used to perceive the scale characteristics of blood vessels after selecting the direction of blood vessels; thereby, adaptively extracting the thick vessel features characterizing the overall characteristics and the thin vessel features preserving the capillaries from fundus images. Then, considering that the single-channel network is unsuitable for the unified characterization of thick and thin vessels, we develop a dual-channel asymmetric CNN based on the U-Net model. The MainSegment-Net uses the step-by-step connection mode to achieve rapid positioning and segmentation of thick vessels; the FineSegment-Net combines dilated convolution and the skip connection to achieve the fine extraction of thin vessels. Finally, the output of the dual-channel asymmetric CNN is fused and coded to combine the segmentation results of thick and thin vessels. The performance of our method is evaluated and tested by DRIVE and CHASE_DB1. The results show that the accuracy (Acc), sensitivity (SE), and specificity (SP) of our method on the DRIVE database are 0.9630, 0.8745, and 0.9823, respectively. The evaluation indexes Acc, SE, and SP of the CHASE_DB1 database are 0.9694, 0.8916, and 0.9794, respectively. Additionally, our method combines the biological vision mechanism with deep learning to achieve rapid and automatic segmentation of retinal vessels, providing a new idea for diagnosing and analyzing subsequent medical images.

The Social Public Issues Analysis Model Based on Deep Learning

To improve the governance effect of public communities and protect community security, combined with the basic principles and network structure of CNN network in deep learning, a community security risk prediction model based on improved single-channel CNN network and decision tree is constructed. The experimental results show that expanding the single channel to the multichannel network greatly extends the receptive field of CNN network. Simultaneously, the accuracy of the prediction model is improved, which is as high as 87.7%. After inputting the output values of every single model into the comprehensive model, the prediction accuracy of the comprehensive model is improved by 4.8%. The feasibility and effectiveness of this method are further demonstrated.

Residual Learning for Marine Mammal Classification

The passive acoustic monitoring of marine mammals is an essential tool for researchers tracking the populations of individual species in threatened environments. Given the large quantity of audio data generated by passive acoustic arrays, it is desirable to automate the process of identifying marine mammals present in the recordings. Utilizing acoustic data from the William A. Watkins Marine Mammal Sounds Database, we present an approach using residual learning networks (ResNets) for classifying the marine mammal vocalizations of up to 32 species. We first determine the optimal methods for converting acoustic recordings into discrete spectrograms suitable for input into neural networks. A series of configurations for spectrographic window functions, preprocessing augmentations, and multi-channel spectrogram generation are examined. Each configuration’s spectrographic output is used to train a residual learning network. Its multi-class classification performance is ranked using the harmonic mean of precision and recall to calculate a weighted F1-score. Configurations specifying 512×256 spectrograms created with a Hann window of 1024 and utilizing horizontal roll demonstrate superior performance. We use the top-performing configurations to generate training data as input for a series of single and multi-channel residual neural networks. These networks are trained to high precision before evaluating their multi-class classification performance. A single-channel network performed the best, obtaining an F1-score of 0.867 with an AUC of 0.9281 on a 32-class classification task. Our multi-channel configuration obtained an F1-score of 0.846 with an AUC of 0.9169. While we demonstrate that networks may learn more information from multi-channel spectrographic inputs, we find that single-channel spectrograms offer superior classification performance overall.

Joint Resource Allocation Optimization of Wireless Sensor Network Based on Edge Computing

Resource allocation has always been a key technology in wireless sensor networks (WSN), but most of the traditional resource allocation algorithms are based on single interface networks. The emergence and development of multi-interface and multichannel networks solve many bottleneck problems of single interface and single channel networks, it also brings new opportunities to the development of wireless sensor networks, but the multi-interface and multichannel technology not only improves the performance of wireless sensor networks but also brings great challenges to the resource allocation of wireless sensor networks. Edge computing changes the traditional centralized cloud computing processing method into a method that reduces computing storage capacity to the edge of the network and faces users and terminals. Realize the advantages of lower latency, higher bandwidth, and fast response. Therefore, this paper proposes a joint optimization algorithm of resource allocation based on edge computing. We establish a wireless sensor allocation model and then propose our algorithm model combined with the advantages of edge computing. Compared with the traditional allocation algorithm (PCOA, MCMH, and TDMA), it can further improve the resource utilization, reduce the network energy consumption, increase network capacity, and reduce the complexity of the schemes.

Subgraph Matching on Multiplex Networks

An active area of research in computational science is the design of algorithms for solving the subgraph matching problem to find copies of a given template graph in a larger world graph. Prior works have largely addressed single-channel networks using a variety of approaches. We present a suite of filtering methods for subgraph isomorphisms for multiplex networks (with different types of edges between nodes and more than one edge within each channel type). We aim to understand the entire solution space rather than focusing on finding one isomorphism. Results are shown on several classes of datasets: (a) Sudoku puzzles mapped to the subgraph isomorphism problem, (b) Erdös-Rényi multigraphs, (c) real-world datasets from Twitter and transportation networks, (d) synthetic data created for the DARPA MAA program.

Machine Learning for Medium Access Control Protocol Recognition in Communications Networks

The ability to recognize the medium access control protocol employed by a network can facilitate the incorporation of a cognitive radio into an existing network by elucidating an integral aspect of network behavior. Since the way in which users access the electromagnetic spectrum is one of the most prominent distinctions between reservation based and contention based medium access control protocols, the first part of this work exploits the regular timing of transmissions from networks utilizing reservation based time-division multiple access (TDMA) protocols to differentiate between transmissions governed by TDMA and by contention based carrier sense multiple access (CSMA) protocols. Our approach leverages modular arithmetic to identify periodicity in transmission timings and an unsupervised $k$ -means algorithm to generate distinct TDMA and CSMA clusters. Several supervised machine learning algorithms are explored to build a protocol classifier. We then present a method of distinguishing between transmissions from multi-channel frequency division multiple access (FDMA) based networks and single channel networks. This method uses an automated machine learning clustering algorithm to obtain an estimate of the actual center frequencies of channels utilized by a network. Such information can be used to determine whether the network is employing an FDMA protocol to access the electromagnetic spectrum.

Galaxy cluster mass estimation with deep learning and hydrodynamical simulations

ABSTRACT We evaluate the ability of convolutional neural networks (CNNs) to predict galaxy cluster masses in the BAHAMAS hydrodynamical simulations. We train four separate single-channel networks using: stellar mass, soft X-ray flux, bolometric X-ray flux, and the Compton y parameter as observational tracers, respectively. Our training set consists of ∼4800 synthetic cluster images generated from the simulation, while an additional ∼3200 images form a validation set and a test set, each with 1600 images. In order to mimic real observation, these images also contain uncorrelated structures located within 50 Mpc in front and behind clusters and seen in projection, as well as instrumental systematics including noise and smoothing. In addition to CNNs for all the four observables, we also train a ‘multichannel’ CNN by combining the four observational tracers. The learning curves of all the five CNNs converge within 1000 epochs. The resulting predictions are especially precise for halo masses in the range $10^{13.25}\, \mathrm{M}_{\odot }\lt M\lt 10^{14.5}\, \mathrm{M}_{\odot }$, where all five networks produce mean mass biases of order ≈1 per cent with a scatter of ≲20 per cent. The network trained with Compton y parameter maps yields the most precise predictions. We interpret the network’s behaviour using two diagnostic tests to determine which features are used to predict cluster mass. The CNNs trained with stellar mass images detect galaxies (not surprisingly), while CNNs trained with gas-based tracers utilize the shape of the signal to estimate cluster mass.

A schedule randomization policy to mitigate timing attacks in WirelessHART networks

Industrial control systems consist of large-scale wireless sensor-actuator networks (WSAN) that control the physical plant. The communications between the sensors and the actuators need to be reliable and guaranteed within strict deadlines for safe operation of all the system components in industrial setups. WirelessHART is the most suitable and widely adopted WSAN standard that serves as the medium of communication in industries. To satisfy deadlines of real-time flows in WirelessHART networks, the centralized network manager decides the communication schedule during network initialization. The same schedule repeats every hyperperiod. The repetitive nature of the communication slots over every hyperperiod makes the system vulnerable to timing attacks which can eventually disrupt the safety of the system. To mitigate such attacks, we propose SlotSwapper, a moving target defense mechanism that randomizes the communication slots over a hyperperiod schedule without violating the feasibility constraints of real-time flows in WirelessHART networks. We show that SlotSwapper is optimal for single-channel WirelessHART network with real-time harmonic flows. We extensively evaluated our algorithm with $$\varvec{4800}$$ flow sets over $$\varvec{100}$$ Tmote sky motes in Cooja simulator. We use Prediction Probability of slots of a schedule to measure the security provided by SlotSwapper. We use Kullback–Leibler divergence to measure the divergence of our solution w.r.t. a truly random solution.

Multi-Channel Fetal ECG Denoising With Deep Convolutional Neural Networks.

Non-invasive fetal electrocardiography represents a valuable alternative continuous fetal monitoring method that has recently received considerable attention in assessing fetal health. However, the non-invasive fetal electrocardiogram (ECG) is typically severely contaminated by a considerable amount of various noise sources, rendering fetal ECG denoising a very challenging task. This work employs a deep learning approach for removing the residual noise from multi-channel fetal ECG after the maternal ECG has been suppressed. We propose a deep convolutional encoder-decoder network with symmetric skip-layer connections, learning end-to-end mappings from noise-corrupted fetal ECG signals to clean ones. Experiments on simulated data show an average signal-to-noise ratio (SNR) improvement of 9.5 dB for fetal ECG signals with input SNR ranging between −20 and 20 dB. The method is additionally evaluated on a large set of real signals, demonstrating that it can provide significant quality improvement of the noisy fetal ECG signals. We further show that employment of multi-channel signal information by the network provides superior and more reliable performance as opposed to its single-channel network counterpart. The presented method is able to preserve beat-to-beat morphological variations and does not require any prior information on the power spectra of the noise or the pulse location.

Steady-State Analysis of the Distributed Queueing Algorithm in a Single-Channel M2M Network

The Distributed Queuing (DQ) algorithm is predicted as one of the solutions to the issues currently found in IoT networks over the use of Aloha based algorithms. Since recently, the algorithm has been of interest to many IoT researchers as a replacement of those Aloha variants for channel access. However, previous works analyzed and evaluated the DQ algorithm without any consideration of the stability of its queues, assuming it is stable for any given number of nodes in the network. In this paper, we define the DQ stability condition in a single-channel M2M environment considering a traffic model of periodic and urgent frames from each node in the network. Besides, a steady-state evaluation of the algorithm’s performance metrics is also presented. In general, the DQ algorithm, when it is stable, was observed not to efficiently use the contention slots for the collision resolution. In a single-channel environment, the DQ algorithm is found to outperform the Aloha based algorithms only in an idle-to-saturation scenario.

Spectrum sensing in single channel and multi-channel cognitive radio networks

<p>Sensing the existence or absence of primary user is the major chore of cognitive radio networks. Nevertheless, Spectrum sensing is the core process of cognitive radio and with target to find idle channels.Various detection techniques exist, however, energy detection is considered as the most used detector because of its lower computational cost. In this paper, we proposed a study of throughput for a cognitive radio system. We had two scenarios, in the first scenario; a study of throughput against probability of false alarm was done; where, only one channel is sensed, to maximize the individual channel throughput. In the second scenario, multi-channel is sensed to maximize the overall system capacity. In addition, different number of channels is considered with different sensing times and at different throughput costs.The performance of the network has been investigated in terms of maximum throughput for optimal number of CR channels. </p>

Stress Field Prediction in Cantilevered Structures Using Convolutional Neural Networks

Abstract The demand for fast and accurate structural analysis is becoming increasingly more prevalent with the advance of generative design and topology optimization technologies. As one step toward accelerating structural analysis, this work explores a deep learning-based approach for predicting the stress fields in 2D linear elastic cantilevered structures subjected to external static loads at its free end using convolutional neural networks (CNNs). Two different architectures are implemented that take as input the structure geometry, external loads, and displacement boundary conditions, and output the predicted von Mises stress field. The first is a single input channel network called SCSNet as the baseline architecture, and the second is the multichannel input network called StressNet. Accuracy analysis shows that StressNet results in significantly lower prediction errors than SCSNet on three loss functions, with a mean relative error of 2.04% for testing. These results suggest that deep learning models may offer a promising alternative to classical methods in structural design and topology optimization. Code and dataset are available.2

Performance comparison of ZRP Bordercasting using Multiple Unicasting vs Broadcasting

MANET (Mobile Ad hoc Networking) is an ad hoc network formed by wireless mobile nodes without any fixed infrastructure. Each node can send, receive and forward data from other nodes. The volatile nature of wireless medium and mobility of nodes pose a great challenge for efficient routing in MANET. There are several routing protocols developed for MANET. ZRP (Zone Routing Protocol) is one of the hybrid MANET routing protocols. It combines proactive and reactive routing approaches for better scalability. In ZRP, the nodes build overlapping zones and maintain topology information of the nodes within their zone. The zone size is decided by zone radius which is defined as no. of hops. Proactive routing is used within the zone and reactive routing is used outside zone. ZRP uses bordercasting to efficiently control the flooding of reactive route queries in outward regions. Each forwarding node propagates the query to selected neighbors which lead to uncovered border nodes of the zone. Ad hoc networks commonly use single channel network. While using single channel network in ZRP, bordercasting can be done either using multiple unicasting or broadcasting. This paper analyses and compares the impact of using multiple unicasting vs broadcasting on routing performance of ZRP, in single channel wireless network. Networking Simulator 2 (NS2) has been used for the simulations. The Qos parameters used for routing performance measurement were Packet delivery ratio (PDR), Average end-to-end delay, throughput and Normalized routing load (NRL). The simulation was done at high mobility by varying network size from small to large network. While using broadcasting, QD2 query control method was used to further control flooding of route queries to uncovered regions. The simulation results conclude that broadcasting is more performance efficient than multiple unicasting for bordercasting in ZRP. General Terms MANET, Mobile Ad hoc networking, ZRP, Zone routing.

Enabling Green Wireless Networking With Device-to-Device Links: A Joint Optimization Approach

Device-to-device (D2D) communication has emerged as a promising technique for improving capacity and reducing power consumption in wireless networks. Most existing works on D2D communications either targeted CDMA-based single-channel networks or aimed at maximizing network throughput. In this paper, we, however, aim to enable green D2D communications in OFDMA-based wireless networks. We formally define an optimization problem based on a practical link data rate model, whose objective is to minimize total power consumption while meeting user data rate requirements. We propose solving it using a joint optimization approach by presenting two effective and efficient algorithms, which both jointly determines mode selection, channel allocation and power assignment. It has been shown by extensive simulation results that the proposed algorithms can achieve over 68% power savings, compared to several baseline methods.

On Capacity and Delay of Multichannel Wireless Networks With Infrastructure Support

In this paper, we propose a novel multi-channel network with infrastructure support, called an MC-IS network, which has not been studied in the literature. To the best of our knowledge, we are the first to study such an MC-IS network. Our proposed MC-IS network has a number of advantages over three existing conventional networks, namely a single-channel wireless ad hoc network (called an SC-AH network), a multi-channel wireless ad hoc network (called an MC-AH network) and a single-channel network with infrastructure support (called an SC-IS network). In particular, the network capacity of our proposed MC-IS network is $\sqrt{n \log n}$ times higher than that of an SC-AH network and an MC-AH network and the same as that of an SC-IS network, where $n$ is the number of nodes in the network. The average delay of our MC-IS network is $\sqrt{\log n/n}$ times lower than that of an SC-AH network and an MC-AH network, and $\min\{C_I,m\}$ times lower than the average delay of an SC-IS network, where $C_I$ and $m$ denote the number of channels dedicated for infrastructure communications and the number of interfaces mounted at each infrastructure node, respectively. Our analysis on an MC-IS network equipped with omni-directional antennas only has been extended to an MC-IS network equipped with directional antennas only, which are named as an MC-IS-DA network. We show that an MC-IS-DA network has an even lower delay of $\frac{c}{\lfloor \frac{2\pi}{\theta}\rfloor \cdot C_I}$ compared with an SC-IS network and our MC-IS network. For example, when $C_I=12$ and $\theta=\frac{\pi}{12}$, an MC-IS-DA network can further reduce the delay by 24 times lower that of an MC-IS network and reduce the delay by 288 times lower than that of an SC-IS network.