Real-Time Speech Enhancement Based on Convolutional Recurrent Neural Network

Speech enhancement is the task of taking a noisy speech input and pro-ducing an enhanced speech output.In recent years,the need for speech enhance-ment has been increased due to challenges that occurred in various applications such as hearing aids,Automatic Speech Recognition(ASR),and mobile speech communication systems.Most of the Speech Enhancement research work has been carried out for English,Chinese,and other European languages.Only a few research works involve speech enhancement in Indian regional Languages.In this paper,we propose a two-fold architecture to perform speech enhancement for Tamil speech signal based on convolutional recurrent neural network(CRN)that addresses the speech enhancement in a real-time single channel or track of sound created by the speaker.In thefirst stage mask based long short-term mem-ory(LSTM)is used for noise suppression along with loss function and in the sec-ond stage,Convolutional Encoder-Decoder(CED)is used for speech restoration.The proposed model is evaluated on various speaker and noisy environments like Babble noise,car noise,and white Gaussian noise.The proposed CRN model improves speech quality by 0.1 points when compared with the LSTM base model and also CRN requires fewer parameters for training.The performance of the pro-posed model is outstanding even in low Signal to Noise Ratio(SNR).

基于ConvLSTM的移动边缘计算服务器能耗模型

针对现有能耗模型对动态工作负载波动具有低敏感性和低精度的问题,该文基于卷积长短期记忆(convolutional long short-term memory, ConvLSTM)神经网络,提出了用于移动边缘计算的服务器智能能耗模型(intelligence server energy consumption model,IECM),用于预测和优化服务器的能量消耗。通过收集服务器运行时间参数,使用熵值法筛选和保留显著影响服务器能耗的参数。基于选定的参数,利用ConvLSTM神经网络训练服务器能耗模型的深度网络。与现有的能耗模型相比,IECM在CPU密集型、I/O密集型、内存密集型和混合型任务上,能够适应服务器工作负载的动态变化,并在能耗预测上具有更好的准确性。

Recurrent Convolutional Neural Network MSER-Based Approach for Payable Document Processing

A tremendous amount of vendor invoices is generated in the corporate sector.To automate the manual data entry in payable documents,highly accurate Optical Character Recognition(OCR)is required.This paper proposes an end-to-end OCR system that does both localization and recognition and serves as a single unit to automate payable document processing such as cheques and cash disbursement.For text localization,the maximally stable extremal region is used,which extracts a word or digit chunk from an invoice.This chunk is later passed to the deep learning model,which performs text recognition.The deep learning model utilizes both convolution neural networks and long short-term memory(LSTM).The convolution layer is used for extracting features,which are fed to the LSTM.The model integrates feature extraction,modeling sequence,and transcription into a unified network.It handles the sequences of unconstrained lengths,independent of the character segmentation or horizontal scale normalization.Furthermore,it applies to both the lexicon-free and lexicon-based text recognition,and finally,it produces a comparatively smaller model,which can be implemented in practical applications.The overall superior performance in the experimental evaluation demonstrates the usefulness of the proposed model.The model is thus generic and can be used for other similar recognition scenarios.

基于CEEMDAN-ConvLSTM组合模型的云计算负载预测方法

随着云计算技术的快速发展,越来越多的用户选择使用云服务。负载请求与资源供应的不匹配问题日益凸显,使得用户请求无法得到及时响应,极大地影响云服务质量,实时预测负载请求,将有助于及时供应资源。针对云计算环境中的负载预测方法性能低的问题,提出了一种基于自适应噪声的完备经验模态分解和卷积长时序神经网络组合模型(CEEMDAN-ConvLSTM)的云计算负载预测方法。首先运用自适应噪声的完备经验模态(CEEMDAN)分解技术对数据序列进行分解操作,将其转换为若干个易于分析和建模的子序列;然后运用卷积长时序神经网络(ConvLSTM)预测模型对这一系列子序列进行建模预测,并采用基于多进程并行计算的研究思路,实现多序列并行预测及贝叶斯优化调参;最后将预测值综合叠加以获得整个模型的预测输出,从而实现对原始复杂序列数据进行高精度预测的目标。使用Google集群工作负载数据集进行实验验证,实验结果表明,CEEMDAN-ConvLSTM组合模型具有良好的预测效果,相比自回归差分移动平均模型(ARIMA)、长短期记忆网络(LSTM)和卷积长时序神经网络(ConvLSTM),所提模型预测均方根误差(RMSE)指标分别提升了30.9%,30.1%和22.5%。

Classification of Arrhythmia Based on Convolutional Neural Networks and Encoder-Decoder Model

As a common and high-risk type of disease,heart disease seriously threatens people’s health.At the same time,in the era of the Internet of Thing(IoT),smart medical device has strong practical significance for medical workers and patients because of its ability to assist in the diagnosis of diseases.Therefore,the research of real-time diagnosis and classification algorithms for arrhythmia can help to improve the diagnostic efficiency of diseases.In this paper,we design an automatic arrhythmia classification algorithm model based on Convolutional Neural Network(CNN)and Encoder-Decoder model.The model uses Long Short-Term Memory(LSTM)to consider the influence of time series features on classification results.Simultaneously,it is trained and tested by the MIT-BIH arrhythmia database.Besides,Generative Adversarial Networks(GAN)is adopted as a method of data equalization for solving data imbalance problem.The simulation results show that for the inter-patient arrhythmia classification,the hybrid model combining CNN and Encoder-Decoder model has the best classification accuracy,of which the accuracy can reach 94.05%.Especially,it has a better advantage for the classification effect of supraventricular ectopic beats(class S)and fusion beats(class F).

Use of Local Region Maps on Convolutional LSTM for Single-Image HDR Reconstruction

Low dynamic range(LDR)images captured by consumer cameras have a limited luminance range.As the conventional method for generating high dynamic range(HDR)images involves merging multiple-exposure LDR images of the same scene(assuming a stationary scene),we introduce a learning-based model for single-image HDR reconstruction.An input LDR image is sequentially segmented into the local region maps based on the cumulative histogram of the input brightness distribution.Using the local region maps,SParam-Net estimates the parameters of an inverse tone mapping function to generate a pseudo-HDR image.We process the segmented region maps as the input sequences on long short-term memory.Finally,a fast super-resolution convolutional neural network is used for HDR image reconstruction.The proposed method was trained and tested on datasets including HDR-Real,LDR-HDR-pair,and HDR-Eye.The experimental results revealed that HDR images can be generated more reliably than using contemporary end-to-end approaches.

Hybrid Model for Short-Term Passenger Flow Prediction in Rail Transit

A precise and timely forecast of short-term rail transit passenger flow provides data support for traffic management and operation,assisting rail operators in efficiently allocating resources and timely relieving pressure on passenger safety and operation.First,the passenger flow sequence models in the study are broken down using VMD for noise reduction.The objective environment features are then added to the characteristic factors that affect the passenger flow.The target station serves as an additional spatial feature and is mined concurrently using the KNN algorithm.It is shown that the hybrid model VMD-CLSMT has a higher prediction accuracy,by setting BP,CNN,and LSTM reference experiments.All models’second order prediction effects are superior to their first order effects,showing that the residual network can significantly raise model prediction accuracy.Additionally,it confirms the efficacy of supplementary and objective environmental features.

Visualization-based prediction of dendritic copper growth in electrochemical cells using convolutional long short-term memory

Electrodeposition in electrochemical cells is one of the leading causes of its performance deterioration. The prediction of electrodeposition growth demands a good understanding of the complex physics involved, which can lead to the fabrication of a probabilistic mathematical model. As an alternative, a convolutional Long shortterm memory architecture-based image analysis approach is presented herein. This technique can predict the electrodeposition growth of the electrolytes, without prior detailed knowledge of the system. The captured images of the electrodeposition from the experiments are used to train and test the model. A comparison between the expected output image and predicted image on a pixel level, percentage mean squared error, absolute percentage error, and pattern density of the electrodeposit are investigated to assess the model accuracy. The randomness of the electrodeposition growth is outlined by investigating the fractal dimension and the interfacial length of the electrodeposits. The trained model predictions show a significant promise between all the experimentally obtained relevant parameters with the predicted one. It is expected that this deep learning-based approach for predicting random electrodeposition growth will be of immense help for designing and optimizing the relevant experimental scheme in near future without performing multiple experiments.

融合时空特征的城市多站点PM2.5浓度预测

本文提出一种融合时空特征的城市多站点PM2.5预测方法,该方法可以捕捉PM2.5在时间和空间上的相关性,通过将区域多个站点的PM2.5数据转换为一系列静态图像,将其输入到卷积长短期记忆(ConvLSTM)模型中,采用端对端的方式进行训练,预测城市未来多个站点多个时段的PM2.5浓度。以北京多个站点的PM2.5数据进行实验验证。结果表明:考虑了时空特征的ConvLSTM方法在短期预测方面优于其他4种时序方法,该方法可为PM2.5预测提供新的思路。

Leucogranite mapping via convolutional recurrent neural networks and geochemical survey data in the Himalayan orogen

Geochemical survey data analysis is recognized as an implemented and feasible way for lithological mapping to assist mineral exploration.With respect to available approaches,recent methodological advances have focused on deep learning algorithms which provide access to learn and extract information directly from geochemical survey data through multi-level networks and outputting end-to-end classification.Accordingly,this study developed a lithological mapping framework with the joint application of a convolutional neural network(CNN)and a long short-term memory(LSTM).The CNN-LSTM model is dominant in correlation extraction from CNN layers and coupling interaction learning from LSTM layers.This hybrid approach was demonstrated by mapping leucogranites in the Himalayan orogen based on stream sediment geochemical survey data,where the targeted leucogranite was expected to be potential resources of rare metals such as Li,Be,and W mineralization.Three comparative case studies were carried out from both visual and quantitative perspectives to illustrate the superiority of the proposed model.A guided spatial distribution map of leucogranites in the Himalayan orogen,divided into high-,moderate-,and low-potential areas,was delineated by the success rate curve,which further improves the efficiency for identifying unmapped leucogranites through geological mapping.In light of these results,this study provides an alternative solution for lithologic mapping using geochemical survey data at a regional scale and reduces the risk for decision making associated with mineral exploration.

Credit Card Fraud Detection Using Improved Deep Learning Models

Fraud of credit cards is a major issue for financial organizations and individuals.As fraudulent actions become more complex,a demand for better fraud detection systems is rising.Deep learning approaches have shown promise in several fields,including detecting credit card fraud.However,the efficacy of these models is heavily dependent on the careful selection of appropriate hyperparameters.This paper introduces models that integrate deep learning models with hyperparameter tuning techniques to learn the patterns and relationships within credit card transaction data,thereby improving fraud detection.Three deep learning models:AutoEncoder(AE),Convolution Neural Network(CNN),and Long Short-Term Memory(LSTM)are proposed to investigate how hyperparameter adjustment impacts the efficacy of deep learning models used to identify credit card fraud.The experiments conducted on a European credit card fraud dataset using different hyperparameters and three deep learning models demonstrate that the proposed models achieve a tradeoff between detection rate and precision,leading these models to be effective in accurately predicting credit card fraud.The results demonstrate that LSTM significantly outperformed AE and CNN in terms of accuracy(99.2%),detection rate(93.3%),and area under the curve(96.3%).These proposed models have surpassed those of existing studies and are expected to make a significant contribution to the field of credit card fraud detection.

Deep Learning for Financial Time Series Prediction:A State-of-the-Art Review of Standalone and HybridModels

Financial time series prediction,whether for classification or regression,has been a heated research topic over the last decade.While traditional machine learning algorithms have experienced mediocre results,deep learning has largely contributed to the elevation of the prediction performance.Currently,the most up-to-date review of advanced machine learning techniques for financial time series prediction is still lacking,making it challenging for finance domain experts and relevant practitioners to determine which model potentially performs better,what techniques and components are involved,and how themodel can be designed and implemented.This review article provides an overview of techniques,components and frameworks for financial time series prediction,with an emphasis on state-of-the-art deep learning models in the literature from2015 to 2023,including standalonemodels like convolutional neural networks(CNN)that are capable of extracting spatial dependencies within data,and long short-term memory(LSTM)that is designed for handling temporal dependencies;and hybrid models integrating CNN,LSTM,attention mechanism(AM)and other techniques.For illustration and comparison purposes,models proposed in recent studies are mapped to relevant elements of a generalized framework comprised of input,output,feature extraction,prediction,and related processes.Among the state-of-the-artmodels,hybrid models like CNNLSTMand CNN-LSTM-AM in general have been reported superior in performance to stand-alone models like the CNN-only model.Some remaining challenges have been discussed,including non-friendliness for finance domain experts,delayed prediction,domain knowledge negligence,lack of standards,and inability of real-time and highfrequency predictions.The principal contributions of this paper are to provide a one-stop guide for both academia and industry to review,compare and summarize technologies and recent advances in this area,to facilitate smooth and informed implementation,and to highlight future research directions.

A multi-source information fusion layer counting method for penetration fuze based on TCN-LSTM

When employing penetration ammunition to strike multi-story buildings,the detection methods using acceleration sensors suffer from signal aliasing,while magnetic detection methods are susceptible to interference from ferromagnetic materials,thereby posing challenges in accurately determining the number of layers.To address this issue,this research proposes a layer counting method for penetration fuze that incorporates multi-source information fusion,utilizing both the temporal convolutional network(TCN)and the long short-term memory(LSTM)recurrent network.By leveraging the strengths of these two network structures,the method extracts temporal and high-dimensional features from the multi-source physical field during the penetration process,establishing a relationship between the multi-source physical field and the distance between the fuze and the target plate.A simulation model is developed to simulate the overload and magnetic field of a projectile penetrating multiple layers of target plates,capturing the multi-source physical field signals and their patterns during the penetration process.The analysis reveals that the proposed multi-source fusion layer counting method reduces errors by 60% and 50% compared to single overload layer counting and single magnetic anomaly signal layer counting,respectively.The model's predictive performance is evaluated under various operating conditions,including different ratios of added noise to random sample positions,penetration speeds,and spacing between target plates.The maximum errors in fuze penetration time predicted by the three modes are 0.08 ms,0.12 ms,and 0.16 ms,respectively,confirming the robustness of the proposed model.Moreover,the model's predictions indicate that the fitting degree for large interlayer spacings is superior to that for small interlayer spacings due to the influence of stress waves.

The Influence of Air Pollution Concentrations on Solar Irradiance Forecasting Using CNN-LSTM-mRMR Feature Extraction

Maintaining a steady power supply requires accurate forecasting of solar irradiance,since clean energy resources do not provide steady power.The existing forecasting studies have examined the limited effects of weather conditions on solar radiation such as temperature and precipitation utilizing convolutional neural network(CNN),but no comprehensive study has been conducted on concentrations of air pollutants along with weather conditions.This paper proposes a hybrid approach based on deep learning,expanding the feature set by adding new air pollution concentrations,and ranking these features to select and reduce their size to improve efficiency.In order to improve the accuracy of feature selection,a maximum-dependency and minimum-redundancy(mRMR)criterion is applied to the constructed feature space to identify and rank the features.The combination of air pollution data with weather conditions data has enabled the prediction of solar irradiance with a higher accuracy.An evaluation of the proposed approach is conducted in Istanbul over 12 months for 43791 discrete times,with the main purpose of analyzing air data,including particular matter(PM10 and PM25),carbon monoxide(CO),nitric oxide(NOX),nitrogen dioxide(NO_(2)),ozone(O₃),sulfur dioxide(SO_(2))using a CNN,a long short-term memory network(LSTM),and MRMR feature extraction.Compared with the benchmark models with root mean square error(RMSE)results of 76.2,60.3,41.3,32.4,there is a significant improvement with the RMSE result of 5.536.This hybrid model presented here offers high prediction accuracy,a wider feature set,and a novel approach based on air concentrations combined with weather conditions for solar irradiance prediction.

A Time Series Intrusion Detection Method Based on SSAE,TCN and Bi-LSTM

In the fast-evolving landscape of digital networks,the incidence of network intrusions has escalated alarmingly.Simultaneously,the crucial role of time series data in intrusion detection remains largely underappreciated,with most systems failing to capture the time-bound nuances of network traffic.This leads to compromised detection accuracy and overlooked temporal patterns.Addressing this gap,we introduce a novel SSAE-TCN-BiLSTM(STL)model that integrates time series analysis,significantly enhancing detection capabilities.Our approach reduces feature dimensionalitywith a Stacked Sparse Autoencoder(SSAE)and extracts temporally relevant features through a Temporal Convolutional Network(TCN)and Bidirectional Long Short-term Memory Network(Bi-LSTM).By meticulously adjusting time steps,we underscore the significance of temporal data in bolstering detection accuracy.On the UNSW-NB15 dataset,ourmodel achieved an F1-score of 99.49%,Accuracy of 99.43%,Precision of 99.38%,Recall of 99.60%,and an inference time of 4.24 s.For the CICDS2017 dataset,we recorded an F1-score of 99.53%,Accuracy of 99.62%,Precision of 99.27%,Recall of 99.79%,and an inference time of 5.72 s.These findings not only confirm the STL model’s superior performance but also its operational efficiency,underpinning its significance in real-world cybersecurity scenarios where rapid response is paramount.Our contribution represents a significant advance in cybersecurity,proposing a model that excels in accuracy and adaptability to the dynamic nature of network traffic,setting a new benchmark for intrusion detection systems.

Track correlation algorithm based on CNN-LSTM for swarm targets

The rapid development of unmanned aerial vehicle(UAV) swarm, a new type of aerial threat target, has brought great pressure to the air defense early warning system. At present, most of the track correlation algorithms only use part of the target location, speed, and other information for correlation.In this paper, the artificial neural network method is used to establish the corresponding intelligent track correlation model and method according to the characteristics of swarm targets.Precisely, a route correlation method based on convolutional neural networks (CNN) and long short-term memory (LSTM)Neural network is designed. In this model, the CNN is used to extract the formation characteristics of UAV swarm and the spatial position characteristics of single UAV track in the formation,while the LSTM is used to extract the time characteristics of UAV swarm. Experimental results show that compared with the traditional algorithms, the algorithm based on CNN-LSTM neural network can make full use of multiple feature information of the target, and has better robustness and accuracy for swarm targets.

Analysis and Modeling of Time Output Characteristics for Distributed Photovoltaic and Energy Storage

Due to the unpredictable output characteristics of distributed photovoltaics,their integration into the grid can lead to voltage fluctuations within the regional power grid.Therefore,the development of spatial-temporal coordination and optimization control methods for distributed photovoltaics and energy storage systems is of utmost importance in various scenarios.This paper approaches the issue from the perspective of spatiotemporal forecasting of distributed photovoltaic(PV)generation and proposes a Temporal Convolutional-Long Short-Term Memory prediction model that combines Temporal Convolutional Networks(TCN)and Long Short-Term Memory(LSTM).To begin with,an analysis of the spatiotemporal distribution patterns of PV generation is conducted,and outlier data is handled using the 3σ rule.Subsequently,a novel approach that combines temporal convolution and LSTM networks is introduced,with TCN extracting spatial features and LSTM capturing temporal features.Finally,a real spatiotemporal dataset from Gansu,China,is established to compare the performance of the proposed network against other models.The results demonstrate that the model presented in this paper exhibits the highest predictive accuracy,with a single-step Mean Absolute Error(MAE)of 1.782 and an average Root Mean Square Error(RMSE)of 3.72 for multi-step predictions.

基于Informer算法的病毒传播预测研究

新冠肺炎病毒等疫情受多种复杂现实因素的影响,因此疫情的发展存在不确定性。为了解决基于传染病仓室模型受自身诸多理想假设条件的限制而导致疫情预测结果误差较大的问题,采用基于深度学习的时序预测模型对疫情发展进行预测,建立了一种基于Transformer模型的Informer模型,并将注意力机制和蒸馏机制应用到疫情数据的时序预测中。以门限自回归(Threshold AutoRegressive, TAR)模型和多种主流的循环神经类时序预测模型作为对比模型,通过仿真实验,对中国、美国和英国的疫情数据当前尚存感染人数进行短期预测,并以均方根误差(RMSE)和平均绝对误差(MAE)为评价指标,选择最佳模型进行了中长期的预测。结果表明,无论是RMSE还是MAE,Informer模型的指标值都是最优的,表明Informer模型对中国、美国和英国疫情的预测精度比其他对比模型高。最后,使用Informer模型对中国、美国和英国的疫情发展进行了中长期预测。

基于CNN-ATT-ConvLSTM的行人属性识别

针对现有行人属性识别方法忽视行人属性的互相关性和空间信息导致识别性能较低的问题,将任务视为时空序列多标签图像分类问题,提出基于卷积神经网络(CNN)和卷积长短期记忆网络(ConvLSTM)并融合通道注意力机制的模型。用CNN和通道注意力提取行人属性的显著性和相关性视觉特征;用ConvLSTM进一步提取视觉特征的空间信息和属性相关性;以优化序列对行人属性进行预测。在两个常用行人属性数据集PETA和RAP上进行大量实验,取得了最佳性能,证明了该方法的优越性和有效性。

Dynamic Hand Gesture Recognition Based on Short-Term Sampling Neural Networks

Hand gestures are a natural way for human-robot interaction.Vision based dynamic hand gesture recognition has become a hot research topic due to its various applications.This paper presents a novel deep learning network for hand gesture recognition.The network integrates several well-proved modules together to learn both short-term and long-term features from video inputs and meanwhile avoid intensive computation.To learn short-term features,each video input is segmented into a fixed number of frame groups.A frame is randomly selected from each group and represented as an RGB image as well as an optical flow snapshot.These two entities are fused and fed into a convolutional neural network(Conv Net)for feature extraction.The Conv Nets for all groups share parameters.To learn longterm features,outputs from all Conv Nets are fed into a long short-term memory(LSTM)network,by which a final classification result is predicted.The new model has been tested with two popular hand gesture datasets,namely the Jester dataset and Nvidia dataset.Comparing with other models,our model produced very competitive results.The robustness of the new model has also been proved with an augmented dataset with enhanced diversity of hand gestures.