A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a force...A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a forced aligned Hidden Markov Model(HMM) state sequence obtained from the GMM-based acoustic model. Therefore, it requires a long computation time for training both the GMM-based acoustic model and a deep learning-based acoustic model. In order to solve this problem, an acoustic model using CTC algorithm is proposed. CTC algorithm does not require the GMM-based acoustic model because it does not use the forced aligned HMM state sequence. However, previous works on a LSTM RNN-based acoustic model using CTC used a small-scale training corpus. In this paper, the LSTM RNN-based acoustic model using CTC is trained on a large-scale training corpus and its performance is evaluated. The implemented acoustic model has a performance of 6.18% and 15.01% in terms of Word Error Rate(WER) for clean speech and noisy speech, respectively. This is similar to a performance of the acoustic model based on the hybrid method.展开更多
A correct and timely fault diagnosis is important for improving the safety and reliability of chemical processes. With the advancement of big data technology, data-driven fault diagnosis methods are being extensively ...A correct and timely fault diagnosis is important for improving the safety and reliability of chemical processes. With the advancement of big data technology, data-driven fault diagnosis methods are being extensively used and still have considerable potential. In recent years, methods based on deep neural networks have made significant breakthroughs, and fault diagnosis methods for industrial processes based on deep learning have attracted considerable research attention. Therefore, we propose a fusion deeplearning algorithm based on a fully convolutional neural network(FCN) to extract features and build models to correctly diagnose all types of faults. We use long short-term memory(LSTM) units to expand our proposed FCN so that our proposed deep learning model can better extract the time-domain features of chemical process data. We also introduce the attention mechanism into the model, aimed at highlighting the importance of features, which is significant for the fault diagnosis of chemical processes with many features. When applied to the benchmark Tennessee Eastman process, our proposed model exhibits impressive performance, demonstrating the effectiveness of the attention-based LSTM FCN in chemical process fault diagnosis.展开更多
There are two technical challenges in predicting slope deformation.The first one is the random displacement,which could not be decomposed and predicted by numerically resolving the observed accumulated displacement an...There are two technical challenges in predicting slope deformation.The first one is the random displacement,which could not be decomposed and predicted by numerically resolving the observed accumulated displacement and time series of a landslide.The second one is the dynamic evolution of a landslide,which could not be feasibly simulated simply by traditional prediction models.In this paper,a dynamic model of displacement prediction is introduced for composite landslides based on a combination of empirical mode decomposition with soft screening stop criteria(SSSC-EMD)and deep bidirectional long short-term memory(DBi-LSTM)neural network.In the proposed model,the time series analysis and SSSC-EMD are used to decompose the observed accumulated displacements of a slope into three components,viz.trend displacement,periodic displacement,and random displacement.Then,by analyzing the evolution pattern of a landslide and its key factors triggering landslides,appropriate influencing factors are selected for each displacement component,and DBi-LSTM neural network to carry out multi-datadriven dynamic prediction for each displacement component.An accumulated displacement prediction has been obtained by a summation of each component.For accuracy verification and engineering practicability of the model,field observations from two known landslides in China,the Xintan landslide and the Bazimen landslide were collected for comparison and evaluation.The case study verified that the model proposed in this paper can better characterize the"stepwise"deformation characteristics of a slope.As compared with long short-term memory(LSTM)neural network,support vector machine(SVM),and autoregressive integrated moving average(ARIMA)model,DBi-LSTM neural network has higher accuracy in predicting the periodic displacement of slope deformation,with the mean absolute percentage error reduced by 3.063%,14.913%,and 13.960%respectively,and the root mean square error reduced by 1.951 mm,8.954 mm and 7.790 mm respectively.Conclusively,this model not only has high prediction accuracy but also is more stable,which can provide new insight for practical landslide prevention and control engineering.展开更多
To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with...To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with sea surface wind and wave heights as training samples.The prediction performance of the model is evaluated,and the error analysis shows that when using the same set of numerically predicted sea surface wind as input,the prediction error produced by the proposed LSTM model at Sta.N01 is 20%,18%and 23%lower than the conventional numerical wave models in terms of the total root mean square error(RMSE),scatter index(SI)and mean absolute error(MAE),respectively.Particularly,for significant wave height in the range of 3–5 m,the prediction accuracy of the LSTM model is improved the most remarkably,with RMSE,SI and MAE all decreasing by 24%.It is also evident that the numbers of hidden neurons,the numbers of buoys used and the time length of training samples all have impact on the prediction accuracy.However,the prediction does not necessary improve with the increase of number of hidden neurons or number of buoys used.The experiment trained by data with the longest time length is found to perform the best overall compared to other experiments with a shorter time length for training.Overall,long short-term memory neural network was proved to be a very promising method for future development and applications in wave forecasting.展开更多
To address the shortcomings of single-step decision making in the existing deep reinforcement learning based unmanned aerial vehicle(UAV)real-time path planning problem,a real-time UAV path planning algorithm based on...To address the shortcomings of single-step decision making in the existing deep reinforcement learning based unmanned aerial vehicle(UAV)real-time path planning problem,a real-time UAV path planning algorithm based on long shortterm memory(RPP-LSTM)network is proposed,which combines the memory characteristics of recurrent neural network(RNN)and the deep reinforcement learning algorithm.LSTM networks are used in this algorithm as Q-value networks for the deep Q network(DQN)algorithm,which makes the decision of the Q-value network has some memory.Thanks to LSTM network,the Q-value network can use the previous environmental information and action information which effectively avoids the problem of single-step decision considering only the current environment.Besides,the algorithm proposes a hierarchical reward and punishment function for the specific problem of UAV real-time path planning,so that the UAV can more reasonably perform path planning.Simulation verification shows that compared with the traditional feed-forward neural network(FNN)based UAV autonomous path planning algorithm,the RPP-LSTM proposed in this paper can adapt to more complex environments and has significantly improved robustness and accuracy when performing UAV real-time path planning.展开更多
Accurate detection of pipeline leakage is essential to maintain the safety of pipeline transportation.Recently,deep learning(DL)has emerged as a promising tool for pipeline leakage detection(PLD).However,most existing...Accurate detection of pipeline leakage is essential to maintain the safety of pipeline transportation.Recently,deep learning(DL)has emerged as a promising tool for pipeline leakage detection(PLD).However,most existing DL methods have difficulty in achieving good performance in identifying leakage types due to the complex time dynamics of pipeline data.On the other hand,the initial parameter selection in the detection model is generally random,which may lead to unstable recognition performance.For this reason,a hybrid DL framework referred to as parameter-optimized recurrent attention network(PRAN)is presented in this paper to improve the accuracy of PLD.First,a parameter-optimized long short-term memory(LSTM)network is introduced to extract effective and robust features,which exploits a particle swarm optimization(PSO)algorithm with cross-entropy fitness function to search for globally optimal parameters.With this framework,the learning representation capability of the model is improved and the convergence rate is accelerated.Moreover,an anomaly-attention mechanism(AM)is proposed to discover class discriminative information by weighting the hidden states,which contributes to amplifying the normalabnormal distinguishable discrepancy,further improving the accuracy of PLD.After that,the proposed PRAN not only implements the adaptive optimization of network parameters,but also enlarges the contribution of normal-abnormal discrepancy,thereby overcoming the drawbacks of instability and poor generalization.Finally,the experimental results demonstrate the effectiveness and superiority of the proposed PRAN for PLD.展开更多
Knowledge of pore-water pressure(PWP)variation is fundamental for slope stability.A precise prediction of PWP is difficult due to complex physical mechanisms and in situ natural variability.To explore the applicabilit...Knowledge of pore-water pressure(PWP)variation is fundamental for slope stability.A precise prediction of PWP is difficult due to complex physical mechanisms and in situ natural variability.To explore the applicability and advantages of recurrent neural networks(RNNs)on PWP prediction,three variants of RNNs,i.e.,standard RNN,long short-term memory(LSTM)and gated recurrent unit(GRU)are adopted and compared with a traditional static artificial neural network(ANN),i.e.,multi-layer perceptron(MLP).Measurements of rainfall and PWP of representative piezometers from a fully instrumented natural slope in Hong Kong are used to establish the prediction models.The coefficient of determination(R^2)and root mean square error(RMSE)are used for model evaluations.The influence of input time series length on the model performance is investigated.The results reveal that MLP can provide acceptable performance but is not robust.The uncertainty bounds of RMSE of the MLP model range from 0.24 kPa to 1.12 k Pa for the selected two piezometers.The standard RNN can perform better but the robustness is slightly affected when there are significant time lags between PWP changes and rainfall.The GRU and LSTM models can provide more precise and robust predictions than the standard RNN.The effects of the hidden layer structure and the dropout technique are investigated.The single-layer GRU is accurate enough for PWP prediction,whereas a double-layer GRU brings extra time cost with little accuracy improvement.The dropout technique is essential to overfitting prevention and improvement of accuracy.展开更多
To supplement missing logging information without increasing economic cost, a machine learning method to generate synthetic well logs from the existing log data was presented, and the experimental verification and app...To supplement missing logging information without increasing economic cost, a machine learning method to generate synthetic well logs from the existing log data was presented, and the experimental verification and application effect analysis were carried out. Since the traditional Fully Connected Neural Network(FCNN) is incapable of preserving spatial dependency, the Long Short-Term Memory(LSTM) network, which is a kind of Recurrent Neural Network(RNN), was utilized to establish a method for log reconstruction. By this method, synthetic logs can be generated from series of input log data with consideration of variation trend and context information with depth. Besides, a cascaded LSTM was proposed by combining the standard LSTM with a cascade system. Testing through real well log data shows that: the results from the LSTM are of higher accuracy than the traditional FCNN; the cascaded LSTM is more suitable for the problem with multiple series data; the machine learning method proposed provides an accurate and cost effective way for synthetic well log generation.展开更多
The remaining useful life(RUL)of a system is generally predicted by utilising the data collected from the sensors that continuously monitor different indicators.Recently,different deep learning(DL)techniques have been...The remaining useful life(RUL)of a system is generally predicted by utilising the data collected from the sensors that continuously monitor different indicators.Recently,different deep learning(DL)techniques have been used for RUL prediction and achieved great success.Because the data is often time-sequential,recurrent neural network(RNN)has attracted significant interests due to its efficiency in dealing with such data.This paper systematically reviews RNN and its variants for RUL prediction,with a specific focus on understanding how different components(e.g.,types of optimisers and activation functions)or parameters(e.g.,sequence length,neuron quantities)affect their performance.After that,a case study using the well-studied NASA’s C-MAPSS dataset is presented to quantitatively evaluate the influence of various state-of-the-art RNN structures on the RUL prediction performance.The result suggests that the variant methods usually perform better than the original RNN,and among which,Bi-directional Long Short-Term Memory generally has the best performance in terms of stability,precision and accuracy.Certain model structures may fail to produce valid RUL prediction result due to the gradient vanishing or gradient exploring problem if the parameters are not chosen appropriately.It is concluded that parameter tuning is a crucial step to achieve optimal prediction performance.展开更多
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 e...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.展开更多
In this paper,we introduce a visual analytics approach aimed at helping machine learning experts analyze the hidden states of layers in recurrent neural networks.Our technique allows the user to interactively inspect ...In this paper,we introduce a visual analytics approach aimed at helping machine learning experts analyze the hidden states of layers in recurrent neural networks.Our technique allows the user to interactively inspect how hidden states store and process information throughout the feeding of an input sequence into the network.The technique can help answer questions,such as which parts of the input data have a higher impact on the prediction and how the model correlates each hidden state configuration with a certain output.Our visual analytics approach comprises several components:First,our input visualization shows the input sequence and how it relates to the output(using color coding).In addition,hidden states are visualized through a nonlinear projection into a 2-D visualization space using t-distributed stochastic neighbor embedding to understand the shape of the space of the hidden states.Trajectories are also employed to show the details of the evolution of the hidden state configurations.Finally,a time-multi-class heatmap matrix visualizes the evolution of the expected predictions for multi-class classifiers,and a histogram indicates the distances between the hidden states within the original space.The different visualizations are shown simultaneously in multiple views and support brushing-and-linking to facilitate the analysis of the classifications and debugging for misclassified input sequences.To demonstrate the capability of our approach,we discuss two typical use cases for long short-term memory models applied to two widely used natural language processing datasets.展开更多
Due to the increase in the types of business and equipment in telecommunications companies,the performance index data collected in the operation and maintenance process varies greatly.The diversity of index data makes...Due to the increase in the types of business and equipment in telecommunications companies,the performance index data collected in the operation and maintenance process varies greatly.The diversity of index data makes it very difficult to perform high-precision capacity prediction.In order to improve the forecasting efficiency of related indexes,this paper designs a classification method of capacity index data,which divides the capacity index data into trend type,periodic type and irregular type.Then for the prediction of trend data,it proposes a capacity index prediction model based on Recurrent Neural Network(RNN),denoted as RNN-LSTM-LSTM.This model includes a basic RNN,two Long Short-Term Memory(LSTM)networks and two Fully Connected layers.The experimental results show that,compared with the traditional Holt-Winters,Autoregressive Integrated Moving Average(ARIMA)and Back Propagation(BP)neural network prediction model,the mean square error(MSE)of the proposed RNN-LSTM-LSTM model are reduced by 11.82%and 20.34%on the order storage and data migration,which has greatly improved the efficiency of trend-type capacity index prediction.展开更多
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 netwo...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.展开更多
Lithium-ion batteries are commonly used in electric vehicles,mobile phones,and laptops.These batteries demonstrate several advantages,such as environmental friendliness,high energy density,and long life.However,batter...Lithium-ion batteries are commonly used in electric vehicles,mobile phones,and laptops.These batteries demonstrate several advantages,such as environmental friendliness,high energy density,and long life.However,battery overcharging and overdischarging may occur if the batteries are not monitored continuously.Overcharging causesfire and explosion casualties,and overdischar-ging causes a reduction in the battery capacity and life.In addition,the internal resistance of such batteries varies depending on their external temperature,elec-trolyte,cathode material,and other factors;the capacity of the batteries decreases with temperature.In this study,we develop a method for estimating the state of charge(SOC)using a neural network model that is best suited to the external tem-perature of such batteries based on their characteristics.During our simulation,we acquired data at temperatures of 25°C,30°C,35°C,and 40°C.Based on the tem-perature parameters,the voltage,current,and time parameters were obtained,and six cycles of the parameters based on the temperature were used for the experi-ment.Experimental data to verify the proposed method were obtained through a discharge experiment conducted using a vehicle driving simulator.The experi-mental data were provided as inputs to three types of neural network models:mul-tilayer neural network(MNN),long short-term memory(LSTM),and gated recurrent unit(GRU).The neural network models were trained and optimized for the specific temperatures measured during the experiment,and the SOC was estimated by selecting the most suitable model for each temperature.The experimental results revealed that the mean absolute errors of the MNN,LSTM,and GRU using the proposed method were 2.17%,2.19%,and 2.15%,respec-tively,which are better than those of the conventional method(4.47%,4.60%,and 4.40%).Finally,SOC estimation based on GRU using the proposed method was found to be 2.15%,which was the most accurate.展开更多
基金supported by the Ministry of Trade,Industry & Energy(MOTIE,Korea) under Industrial Technology Innovation Program (No.10063424,'development of distant speech recognition and multi-task dialog processing technologies for in-door conversational robots')
文摘A Long Short-Term Memory(LSTM) Recurrent Neural Network(RNN) has driven tremendous improvements on an acoustic model based on Gaussian Mixture Model(GMM). However, these models based on a hybrid method require a forced aligned Hidden Markov Model(HMM) state sequence obtained from the GMM-based acoustic model. Therefore, it requires a long computation time for training both the GMM-based acoustic model and a deep learning-based acoustic model. In order to solve this problem, an acoustic model using CTC algorithm is proposed. CTC algorithm does not require the GMM-based acoustic model because it does not use the forced aligned HMM state sequence. However, previous works on a LSTM RNN-based acoustic model using CTC used a small-scale training corpus. In this paper, the LSTM RNN-based acoustic model using CTC is trained on a large-scale training corpus and its performance is evaluated. The implemented acoustic model has a performance of 6.18% and 15.01% in terms of Word Error Rate(WER) for clean speech and noisy speech, respectively. This is similar to a performance of the acoustic model based on the hybrid method.
文摘A correct and timely fault diagnosis is important for improving the safety and reliability of chemical processes. With the advancement of big data technology, data-driven fault diagnosis methods are being extensively used and still have considerable potential. In recent years, methods based on deep neural networks have made significant breakthroughs, and fault diagnosis methods for industrial processes based on deep learning have attracted considerable research attention. Therefore, we propose a fusion deeplearning algorithm based on a fully convolutional neural network(FCN) to extract features and build models to correctly diagnose all types of faults. We use long short-term memory(LSTM) units to expand our proposed FCN so that our proposed deep learning model can better extract the time-domain features of chemical process data. We also introduce the attention mechanism into the model, aimed at highlighting the importance of features, which is significant for the fault diagnosis of chemical processes with many features. When applied to the benchmark Tennessee Eastman process, our proposed model exhibits impressive performance, demonstrating the effectiveness of the attention-based LSTM FCN in chemical process fault diagnosis.
文摘There are two technical challenges in predicting slope deformation.The first one is the random displacement,which could not be decomposed and predicted by numerically resolving the observed accumulated displacement and time series of a landslide.The second one is the dynamic evolution of a landslide,which could not be feasibly simulated simply by traditional prediction models.In this paper,a dynamic model of displacement prediction is introduced for composite landslides based on a combination of empirical mode decomposition with soft screening stop criteria(SSSC-EMD)and deep bidirectional long short-term memory(DBi-LSTM)neural network.In the proposed model,the time series analysis and SSSC-EMD are used to decompose the observed accumulated displacements of a slope into three components,viz.trend displacement,periodic displacement,and random displacement.Then,by analyzing the evolution pattern of a landslide and its key factors triggering landslides,appropriate influencing factors are selected for each displacement component,and DBi-LSTM neural network to carry out multi-datadriven dynamic prediction for each displacement component.An accumulated displacement prediction has been obtained by a summation of each component.For accuracy verification and engineering practicability of the model,field observations from two known landslides in China,the Xintan landslide and the Bazimen landslide were collected for comparison and evaluation.The case study verified that the model proposed in this paper can better characterize the"stepwise"deformation characteristics of a slope.As compared with long short-term memory(LSTM)neural network,support vector machine(SVM),and autoregressive integrated moving average(ARIMA)model,DBi-LSTM neural network has higher accuracy in predicting the periodic displacement of slope deformation,with the mean absolute percentage error reduced by 3.063%,14.913%,and 13.960%respectively,and the root mean square error reduced by 1.951 mm,8.954 mm and 7.790 mm respectively.Conclusively,this model not only has high prediction accuracy but also is more stable,which can provide new insight for practical landslide prevention and control engineering.
基金The National Key R&D Program of China under contract No.2016YFC1402103
文摘To explore new operational forecasting methods of waves,a forecasting model for wave heights at three stations in the Bohai Sea has been developed.This model is based on long short-term memory(LSTM)neural network with sea surface wind and wave heights as training samples.The prediction performance of the model is evaluated,and the error analysis shows that when using the same set of numerically predicted sea surface wind as input,the prediction error produced by the proposed LSTM model at Sta.N01 is 20%,18%and 23%lower than the conventional numerical wave models in terms of the total root mean square error(RMSE),scatter index(SI)and mean absolute error(MAE),respectively.Particularly,for significant wave height in the range of 3–5 m,the prediction accuracy of the LSTM model is improved the most remarkably,with RMSE,SI and MAE all decreasing by 24%.It is also evident that the numbers of hidden neurons,the numbers of buoys used and the time length of training samples all have impact on the prediction accuracy.However,the prediction does not necessary improve with the increase of number of hidden neurons or number of buoys used.The experiment trained by data with the longest time length is found to perform the best overall compared to other experiments with a shorter time length for training.Overall,long short-term memory neural network was proved to be a very promising method for future development and applications in wave forecasting.
基金supported by the Natural Science Basic Research Prog ram of Shaanxi(2022JQ-593)。
文摘To address the shortcomings of single-step decision making in the existing deep reinforcement learning based unmanned aerial vehicle(UAV)real-time path planning problem,a real-time UAV path planning algorithm based on long shortterm memory(RPP-LSTM)network is proposed,which combines the memory characteristics of recurrent neural network(RNN)and the deep reinforcement learning algorithm.LSTM networks are used in this algorithm as Q-value networks for the deep Q network(DQN)algorithm,which makes the decision of the Q-value network has some memory.Thanks to LSTM network,the Q-value network can use the previous environmental information and action information which effectively avoids the problem of single-step decision considering only the current environment.Besides,the algorithm proposes a hierarchical reward and punishment function for the specific problem of UAV real-time path planning,so that the UAV can more reasonably perform path planning.Simulation verification shows that compared with the traditional feed-forward neural network(FNN)based UAV autonomous path planning algorithm,the RPP-LSTM proposed in this paper can adapt to more complex environments and has significantly improved robustness and accuracy when performing UAV real-time path planning.
基金This work was supported in part by the National Natural Science Foundation of China(U21A2019,61873058),Hainan Province Science and Technology Special Fund of China(ZDYF2022SHFZ105)the Alexander von Humboldt Foundation of Germany.
文摘Accurate detection of pipeline leakage is essential to maintain the safety of pipeline transportation.Recently,deep learning(DL)has emerged as a promising tool for pipeline leakage detection(PLD).However,most existing DL methods have difficulty in achieving good performance in identifying leakage types due to the complex time dynamics of pipeline data.On the other hand,the initial parameter selection in the detection model is generally random,which may lead to unstable recognition performance.For this reason,a hybrid DL framework referred to as parameter-optimized recurrent attention network(PRAN)is presented in this paper to improve the accuracy of PLD.First,a parameter-optimized long short-term memory(LSTM)network is introduced to extract effective and robust features,which exploits a particle swarm optimization(PSO)algorithm with cross-entropy fitness function to search for globally optimal parameters.With this framework,the learning representation capability of the model is improved and the convergence rate is accelerated.Moreover,an anomaly-attention mechanism(AM)is proposed to discover class discriminative information by weighting the hidden states,which contributes to amplifying the normalabnormal distinguishable discrepancy,further improving the accuracy of PLD.After that,the proposed PRAN not only implements the adaptive optimization of network parameters,but also enlarges the contribution of normal-abnormal discrepancy,thereby overcoming the drawbacks of instability and poor generalization.Finally,the experimental results demonstrate the effectiveness and superiority of the proposed PRAN for PLD.
基金supported by the Natural Science Foundation of China(Grant Nos.51979158,51639008,51679135,and 51422905)the Program of Shanghai Academic Research Leader by Science and Technology Commission of Shanghai Municipality(Project No.19XD1421900)。
文摘Knowledge of pore-water pressure(PWP)variation is fundamental for slope stability.A precise prediction of PWP is difficult due to complex physical mechanisms and in situ natural variability.To explore the applicability and advantages of recurrent neural networks(RNNs)on PWP prediction,three variants of RNNs,i.e.,standard RNN,long short-term memory(LSTM)and gated recurrent unit(GRU)are adopted and compared with a traditional static artificial neural network(ANN),i.e.,multi-layer perceptron(MLP).Measurements of rainfall and PWP of representative piezometers from a fully instrumented natural slope in Hong Kong are used to establish the prediction models.The coefficient of determination(R^2)and root mean square error(RMSE)are used for model evaluations.The influence of input time series length on the model performance is investigated.The results reveal that MLP can provide acceptable performance but is not robust.The uncertainty bounds of RMSE of the MLP model range from 0.24 kPa to 1.12 k Pa for the selected two piezometers.The standard RNN can perform better but the robustness is slightly affected when there are significant time lags between PWP changes and rainfall.The GRU and LSTM models can provide more precise and robust predictions than the standard RNN.The effects of the hidden layer structure and the dropout technique are investigated.The single-layer GRU is accurate enough for PWP prediction,whereas a double-layer GRU brings extra time cost with little accuracy improvement.The dropout technique is essential to overfitting prevention and improvement of accuracy.
基金Supported by the National Natural Science Foundation of China(U1663208,51520105005)the National Science and Technology Major Project of China(2017ZX05009-005,2016ZX05037-003)
文摘To supplement missing logging information without increasing economic cost, a machine learning method to generate synthetic well logs from the existing log data was presented, and the experimental verification and application effect analysis were carried out. Since the traditional Fully Connected Neural Network(FCNN) is incapable of preserving spatial dependency, the Long Short-Term Memory(LSTM) network, which is a kind of Recurrent Neural Network(RNN), was utilized to establish a method for log reconstruction. By this method, synthetic logs can be generated from series of input log data with consideration of variation trend and context information with depth. Besides, a cascaded LSTM was proposed by combining the standard LSTM with a cascade system. Testing through real well log data shows that: the results from the LSTM are of higher accuracy than the traditional FCNN; the cascaded LSTM is more suitable for the problem with multiple series data; the machine learning method proposed provides an accurate and cost effective way for synthetic well log generation.
基金Supported by U.K.EPSRC Platform Grant(Grant No.EP/P027121/1).
文摘The remaining useful life(RUL)of a system is generally predicted by utilising the data collected from the sensors that continuously monitor different indicators.Recently,different deep learning(DL)techniques have been used for RUL prediction and achieved great success.Because the data is often time-sequential,recurrent neural network(RNN)has attracted significant interests due to its efficiency in dealing with such data.This paper systematically reviews RNN and its variants for RUL prediction,with a specific focus on understanding how different components(e.g.,types of optimisers and activation functions)or parameters(e.g.,sequence length,neuron quantities)affect their performance.After that,a case study using the well-studied NASA’s C-MAPSS dataset is presented to quantitatively evaluate the influence of various state-of-the-art RNN structures on the RUL prediction performance.The result suggests that the variant methods usually perform better than the original RNN,and among which,Bi-directional Long Short-Term Memory generally has the best performance in terms of stability,precision and accuracy.Certain model structures may fail to produce valid RUL prediction result due to the gradient vanishing or gradient exploring problem if the parameters are not chosen appropriately.It is concluded that parameter tuning is a crucial step to achieve optimal prediction performance.
基金Researchers would like to thank the Deanship of Scientific Research,Qassim University,for funding publication of this project.
文摘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.
基金Funded by the Deutsche Forschungsgemeinschaft(German Research Foundation),No.251654672—TRR 161(Project B01)Germany’s Excellence Strategy,No.EXC-2075—390740016.
文摘In this paper,we introduce a visual analytics approach aimed at helping machine learning experts analyze the hidden states of layers in recurrent neural networks.Our technique allows the user to interactively inspect how hidden states store and process information throughout the feeding of an input sequence into the network.The technique can help answer questions,such as which parts of the input data have a higher impact on the prediction and how the model correlates each hidden state configuration with a certain output.Our visual analytics approach comprises several components:First,our input visualization shows the input sequence and how it relates to the output(using color coding).In addition,hidden states are visualized through a nonlinear projection into a 2-D visualization space using t-distributed stochastic neighbor embedding to understand the shape of the space of the hidden states.Trajectories are also employed to show the details of the evolution of the hidden state configurations.Finally,a time-multi-class heatmap matrix visualizes the evolution of the expected predictions for multi-class classifiers,and a histogram indicates the distances between the hidden states within the original space.The different visualizations are shown simultaneously in multiple views and support brushing-and-linking to facilitate the analysis of the classifications and debugging for misclassified input sequences.To demonstrate the capability of our approach,we discuss two typical use cases for long short-term memory models applied to two widely used natural language processing datasets.
基金supported by Research on Big Data Technology for New Generation Internet Operators(H04W180609)the second batch of Sichuan Science and Technology Service Industry Development Fund Projects in 2018(18KJFWSF0388).
文摘Due to the increase in the types of business and equipment in telecommunications companies,the performance index data collected in the operation and maintenance process varies greatly.The diversity of index data makes it very difficult to perform high-precision capacity prediction.In order to improve the forecasting efficiency of related indexes,this paper designs a classification method of capacity index data,which divides the capacity index data into trend type,periodic type and irregular type.Then for the prediction of trend data,it proposes a capacity index prediction model based on Recurrent Neural Network(RNN),denoted as RNN-LSTM-LSTM.This model includes a basic RNN,two Long Short-Term Memory(LSTM)networks and two Fully Connected layers.The experimental results show that,compared with the traditional Holt-Winters,Autoregressive Integrated Moving Average(ARIMA)and Back Propagation(BP)neural network prediction model,the mean square error(MSE)of the proposed RNN-LSTM-LSTM model are reduced by 11.82%and 20.34%on the order storage and data migration,which has greatly improved the efficiency of trend-type capacity index prediction.
文摘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.
基金supported by the BK21 FOUR project funded by the Ministry of Education,Korea(4199990113966).
文摘Lithium-ion batteries are commonly used in electric vehicles,mobile phones,and laptops.These batteries demonstrate several advantages,such as environmental friendliness,high energy density,and long life.However,battery overcharging and overdischarging may occur if the batteries are not monitored continuously.Overcharging causesfire and explosion casualties,and overdischar-ging causes a reduction in the battery capacity and life.In addition,the internal resistance of such batteries varies depending on their external temperature,elec-trolyte,cathode material,and other factors;the capacity of the batteries decreases with temperature.In this study,we develop a method for estimating the state of charge(SOC)using a neural network model that is best suited to the external tem-perature of such batteries based on their characteristics.During our simulation,we acquired data at temperatures of 25°C,30°C,35°C,and 40°C.Based on the tem-perature parameters,the voltage,current,and time parameters were obtained,and six cycles of the parameters based on the temperature were used for the experi-ment.Experimental data to verify the proposed method were obtained through a discharge experiment conducted using a vehicle driving simulator.The experi-mental data were provided as inputs to three types of neural network models:mul-tilayer neural network(MNN),long short-term memory(LSTM),and gated recurrent unit(GRU).The neural network models were trained and optimized for the specific temperatures measured during the experiment,and the SOC was estimated by selecting the most suitable model for each temperature.The experimental results revealed that the mean absolute errors of the MNN,LSTM,and GRU using the proposed method were 2.17%,2.19%,and 2.15%,respec-tively,which are better than those of the conventional method(4.47%,4.60%,and 4.40%).Finally,SOC estimation based on GRU using the proposed method was found to be 2.15%,which was the most accurate.