Real-time perception of rock mass information is of great importance to efficient tunneling and hazard prevention in tunnel boring machines(TBMs).In this study,a TBM-rock mutual feedback perception method based on dat...Real-time perception of rock mass information is of great importance to efficient tunneling and hazard prevention in tunnel boring machines(TBMs).In this study,a TBM-rock mutual feedback perception method based on data mining(DM) is proposed,which takes 10 tunneling parameters related to surrounding rock conditions as input features.For implementation,first,the database of TBM tunneling parameters was established,in which 10,807 tunneling cycles from the Songhua River water conveyance tunnel were accommodated.Then,the spectral clustering(SC) algorithm based on graph theory was introduced to cluster the TBM tunneling data.According to the clustering results and rock mass boreability index,the rock mass conditions were classified into four classes,and the reasonable distribution intervals of the main tunneling parameters corresponding to each class were presented.Meanwhile,based on the deep neural network(DNN),the real-time prediction model regarding different rock conditions was established.Finally,the rationality and adaptability of the proposed method were validated via analyzing the tunneling specific energy,feature importance,and training dataset size.The proposed TBM-rock mutual feedback perception method enables the automatic identification of rock mass conditions and the dynamic adjustment of tunneling parameters during TBM driving.Furthermore,in terms of the prediction performance,the method can predict the rock mass conditions ahead of the tunnel face in real time more accurately than the traditional machine learning prediction methods.展开更多
In the past decade,dramatic progress has been made in the field of machine learning.This paper explores the possibility of applying deep learning in power system state estimation.Traditionally,physics-based models are...In the past decade,dramatic progress has been made in the field of machine learning.This paper explores the possibility of applying deep learning in power system state estimation.Traditionally,physics-based models are used including weighted least square(WLS)or weighted least absolute value(WLAV).These models typically consider a single snapshot of the system without capturing temporal correlations of system states.In this paper,a physics-guided deep learning(PGDL)method is proposed.Specifically,inspired by autoencoders,deep neural networks(DNNs)are used to learn the temporal correlations.The estimated system states from DNNs are then checked against physics laws by running through a set of power flow equations.Hence,the proposed PGDL is both data-driven and physics-guided.The accuracy and robustness of the proposed PGDL method are compared with traditional methods in standard IEEE cases.Simulations show promising results and the applicability is further discussed.展开更多
For training the present Neural Network(NN)models,the standard technique is to utilize decaying Learning Rates(LR).While the majority of these techniques commence with a large LR,they will decay multiple times over ti...For training the present Neural Network(NN)models,the standard technique is to utilize decaying Learning Rates(LR).While the majority of these techniques commence with a large LR,they will decay multiple times over time.Decaying has been proved to enhance generalization as well as optimization.Other parameters,such as the network’s size,the number of hidden layers,drop-outs to avoid overfitting,batch size,and so on,are solely based on heuristics.This work has proposed Adaptive Teaching Learning Based(ATLB)Heuristic to identify the optimal hyperparameters for diverse networks.Here we consider three architec-tures Recurrent Neural Networks(RNN),Long Short Term Memory(LSTM),Bidirectional Long Short Term Memory(BiLSTM)of Deep Neural Networks for classification.The evaluation of the proposed ATLB is done through the various learning rate schedulers Cyclical Learning Rate(CLR),Hyperbolic Tangent Decay(HTD),and Toggle between Hyperbolic Tangent Decay and Triangular mode with Restarts(T-HTR)techniques.Experimental results have shown the performance improvement on the 20Newsgroup,Reuters Newswire and IMDB dataset.展开更多
We present a numerical approach for modeling unknown dynamical systems using partially observed data,with a focus on biological systems with(relatively)complex dynamical behavior.As an extension of the recently develo...We present a numerical approach for modeling unknown dynamical systems using partially observed data,with a focus on biological systems with(relatively)complex dynamical behavior.As an extension of the recently developed deep neural network(DNN)learning methods,our approach is particularly suitable for practical situations when(i)measurement data are available for only a subset of the state variables,and(ii)the system parameters cannot be observed or measured at all.We demonstrate that,with a properly designed DNN structure with memory terms,effective DNN models can be learned from such partially observed data containing hidden parameters.The learned DNN model serves as an accurate predictive tool for system analysis.Through a few representative biological problems,we demonstrate that such DNN models can capture qualitative dynamical behavior changes in the system,such as bifurcations,even when the parameters controlling such behavior changes are completely unknown throughout not only the model learning process but also the system prediction process.The learned DNN model effectively creates a“closed”model involving only the observables when such a closed-form model does not exist mathematically.展开更多
Reinforcement learning(RL)has become a dominant decision-making paradigm and has achieved notable success in many real-world applications.Notably,deep neural networks play a crucial role in unlocking RL’s potential i...Reinforcement learning(RL)has become a dominant decision-making paradigm and has achieved notable success in many real-world applications.Notably,deep neural networks play a crucial role in unlocking RL’s potential in large-scale decision-making tasks.Inspired by current major success of Transformer in natural language processing and computer vision,numerous bottlenecks have been overcome by combining Transformer with RL for decision-making.This paper presents a multiangle systematic survey of various Transformer-based RL(TransRL)models applied in decision-making tasks,including basic models,advanced algorithms,representative implementation instances,typical applications,and known challenges.Our work aims to provide insights into problems that inherently arise with the current RL approaches,and examines how we can address them with better TransRL models.To our knowledge,we are the first to present a comprehensive review of the recent Transformer research developments in RL for decision-making.We hope that this survey provides a comprehensive review of TransRL models and inspires the RL community in its pursuit of future directions.To keep track of the rapid TransRL developments in the decision-making domains,we summarize the latest papers and their open-source implementations at https://github.com/williamyuanv0/Transformer-in-Reinforcement-Learning-for-Decision-Making-A-Survey.展开更多
Sheet metal forming technologies have been intensively studied for decades to meet the increasing demand for lightweight metal components.To surmount the springback occurring in sheet metal forming processes,numerous ...Sheet metal forming technologies have been intensively studied for decades to meet the increasing demand for lightweight metal components.To surmount the springback occurring in sheet metal forming processes,numerous studies have been performed to develop compensation methods.However,for most existing methods,the development cycle is still considerably time-consumptive and demands high computational or capital cost.In this paper,a novel theory-guided regularization method for training of deep neural networks(DNNs),implanted in a learning system,is introduced to learn the intrinsic relationship between the workpiece shape after springback and the required process parameter,e.g.,loading stroke,in sheet metal bending processes.By directly bridging the workpiece shape to the process parameter,issues concerning springback in the process design would be circumvented.The novel regularization method utilizes the well-recognized theories in material mechanics,Swift’s law,by penalizing divergence from this law throughout the network training process.The regularization is implemented by a multi-task learning network architecture,with the learning of extra tasks regularized during training.The stress-strain curve describing the material properties and the prior knowledge used to guide learning are stored in the database and the knowledge base,respectively.One can obtain the predicted loading stroke for a new workpiece shape by importing the target geometry through the user interface.In this research,the neural models were found to outperform a traditional machine learning model,support vector regression model,in experiments with different amount of training data.Through a series of studies with varying conditions of training data structure and amount,workpiece material and applied bending processes,the theory-guided DNN has been shown to achieve superior generalization and learning consistency than the data-driven DNNs,especially when only scarce and scattered experiment data are available for training which is often the c展开更多
In industrial process control systems,there is overwhelming evidence corroborating the notion that economic or technical limitations result in some key variables that are very difficult to measure online.The data-driv...In industrial process control systems,there is overwhelming evidence corroborating the notion that economic or technical limitations result in some key variables that are very difficult to measure online.The data-driven soft sensor is an effective solution because it provides a reliable and stable online estimation of such variables.This paper employs a deep neural network with multiscale feature extraction layers to build soft sensors,which are applied to the benchmarked Tennessee-Eastman process(TEP)and a real wind farm case.The comparison of modelling results demonstrates that the multiscale feature extraction layers have the following advantages over other methods.First,the multiscale feature extraction layers significantly reduce the number of parameters compared to the other deep neural networks.Second,the multiscale feature extraction layers can powerfully extract dataset characteristics.Finally,the multiscale feature extraction layers with fully considered historical measurements can contain richer useful information and improved representation compared to traditional data-driven models.展开更多
With the vigorous development of artificial intelligence(AI),intelligence applications based on deep neural networks(DNNs)have changed people’s lifestyles and production efficiency.However,the large amount of computa...With the vigorous development of artificial intelligence(AI),intelligence applications based on deep neural networks(DNNs)have changed people’s lifestyles and production efficiency.However,the large amount of computation and data generated from the network edge becomes the major bottleneck,and the traditional cloud-based computing mode has been unable to meet the requirements of realtime processing tasks.To solve the above problems,by embedding AI model training and inference capabilities into the network edge,edge intelligence(EI)becomes a cutting-edge direction in the field of AI.Furthermore,collaborative DNN inference among the cloud,edge,and end devices provides a promising way to boost EI.Nevertheless,at present,EI oriented collaborative DNN inference is still in its early stage,lacking systematic classification and discussion of existing research efforts.Motivated by it,we have comprehensively investigated recent studies on EI-oriented collaborative DNN inference.In this paper,we first review the background and motivation of EI.Then,we classify four typical collaborative DNN inference paradigms for EI,and analyse their characteristics and key technologies.Finally,we summarize the current challenges of collaborative DNN inference,discuss future development trends and provide future research directions.展开更多
The access of unified power flow controllers(UPFC)has changed the structure and operation mode of power grids all across the world,and it has brought severe challenges to the traditional real-time calculation of secur...The access of unified power flow controllers(UPFC)has changed the structure and operation mode of power grids all across the world,and it has brought severe challenges to the traditional real-time calculation of security correction based on traditionalmodels.Considering the limitation of computational efficiency regarding complex,physical models,a data-driven power system security correction method with UPFC is,in this paper,proposed.Based on the complex mapping relationship between the operation state data and the security correction strategy,a two-stage deep neural network(DNN)learning framework is proposed,which divides the offline training task of security correction into two stages:in the first stage,the stacked auto-encoder(SAE)classification model is established,and the node correction state(0/1)output based on the fault information;in the second stage,the DNN learningmodel is established,and the correction amount of each action node is obtained based on the action nodes output in the previous stage.In this paper,the UPFC demonstration project of NanjingWest Ring Network is taken as a case study to validate the proposed method.The results show that the proposed method can fully meet the real-time security correction time requirements of power grids,and avoid the inherent defects of the traditional model method without an iterative solution and can also provide reasonable security correction strategies for N-1 and N-2 faults.展开更多
Real-time guidance is critical for the vertical recovery of rockets.However,traditional sequential convex optimization algorithms suffer from shortcomings in terms of their poor real-time performance.This work focuses...Real-time guidance is critical for the vertical recovery of rockets.However,traditional sequential convex optimization algorithms suffer from shortcomings in terms of their poor real-time performance.This work focuses on applying the deep learning-based closedloop guidance algorithm and error propagation analysis for powered landing,thereby significantly improving the real-time performance.First,a controller consisting of two deep neural networks is constructed to map the thrust direction and magnitude of the rocket according to the state variables.Thereafter,the analytical transition relationships between different uncertainty sources and the state propagation error in a single guidance period are analyzed by adopting linear covariance analysis.Finally,the accuracy of the proposed methods is verified via a comparison with the indirect method and Monte Carlo simulations.Compared with the traditional sequential convex optimization algorithm,our method reduces the computation time from 75 ms to less than 1 ms.Therefore,it shows potential for online applications.展开更多
This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)mod...This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)model,then a welldesigned RL algorithm,experience based deep deterministic policy gradient(EBDDPG),is proposed to solve it.By taking the advantage of prior information generated through the optimal control model,the proposed algorithm not only resolves the convergence problem of the common RL algorithm,but also successfully trains an efficient deep neural network(DNN)controller for the chaser spacecraft to generate the control sequence.Numerical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.展开更多
DNA methylation analysis has been applied to determine the primary site of cancer;however, robust and accurate prediction of cancer types with a minimum number of sites is still a significant scientific challenge. To ...DNA methylation analysis has been applied to determine the primary site of cancer;however, robust and accurate prediction of cancer types with a minimum number of sites is still a significant scientific challenge. To build an accurate and robust cancer type prediction tool with a minimum number of DNA methylation sites, we internally benchmarked different DNA methylation site selection and ranking procedures, as well as different classification models. We used The Cancer Genome Atlas dataset (26 cancer types with 8296 samples) to train and test models and used an independent dataset (17 cancer types with 2738 samples) for model validation. A deep neural network model using a combined feature selection procedure (named MethyDeep) can predict 26 cancer types using 30 methylation sites with superior performance compared with the known methods for both primary and metastatic cancers in independent validation datasets. In conclusion, MethyDeep is an accurate and robust cancer type predictor with the minimum number of DNA methylation sites;it could help the cost-effective clarification of cancer of unknown primary patients and the liquid biopsy-based early screening of cancers.展开更多
Distributed denial of service(DDoS)attack is the most common attack that obstructs a network and makes it unavailable for a legitimate user.We proposed a deep neural network(DNN)model for the detection of DDoS attacks...Distributed denial of service(DDoS)attack is the most common attack that obstructs a network and makes it unavailable for a legitimate user.We proposed a deep neural network(DNN)model for the detection of DDoS attacks in the Software-Defined Networking(SDN)paradigm.SDN centralizes the control plane and separates it from the data plane.It simplifies a network and eliminates vendor specification of a device.Because of this open nature and centralized control,SDN can easily become a victim of DDoS attacks.We proposed a supervised Developed Deep Neural Network(DDNN)model that can classify the DDoS attack traffic and legitimate traffic.Our Developed Deep Neural Network(DDNN)model takes a large number of feature values as compared to previously proposed Machine Learning(ML)models.The proposed DNN model scans the data to find the correlated features and delivers high-quality results.The model enhances the security of SDN and has better accuracy as compared to previously proposed models.We choose the latest state-of-the-art dataset which consists of many novel attacks and overcomes all the shortcomings and limitations of the existing datasets.Our model results in a high accuracy rate of 99.76%with a low false-positive rate and 0.065%low loss rate.The accuracy increases to 99.80%as we increase the number of epochs to 100 rounds.Our proposed model classifies anomalous and normal traffic more accurately as compared to the previously proposed models.It can handle a huge amount of structured and unstructured data and can easily solve complex problems.展开更多
Video Description aims to automatically generate descriptive natural language for videos.Due to the large volume of multi-modal data and successful implementations of Deep Neural Networks(DNNs),a wide range of models ...Video Description aims to automatically generate descriptive natural language for videos.Due to the large volume of multi-modal data and successful implementations of Deep Neural Networks(DNNs),a wide range of models have been proposed.However,previous models learn insufficient linguistic information or correlation between visual and textual modalities.In order to address those problems,this paper proposes an integrated model using Long Short-Term Memory(LSTM).This proposed model consists of triple channels in parallel:a primary video description channel,a sentence-to-sentence channel for language learning,and a channel to integrate visual and textual information.Additionally,the parallel three channels are connected by LSTM weight matrices during training.The VD-ivt model is evaluated on two publicly available datasets,i.e.Youtube2 Text and LSMDC.Experimental results demonstrate that the performance of the proposed model outperforms those benchmarks.展开更多
The literature generally agrees that longer-horizon(over a month) predictions make more sense than short-horizon ones. However, it's an especially challenging task due to the lack of data(in unit of long horizon)a...The literature generally agrees that longer-horizon(over a month) predictions make more sense than short-horizon ones. However, it's an especially challenging task due to the lack of data(in unit of long horizon)and economic data have a low S/N ratio. We hypothesize that the stock trend is largely dictated by driving factors which are filtered by psychological factors and work on behavioral factors: representative indicators from these three aspects would be adequate in trend prediction. We then extend the Stepwise Regression Analysis(SRA)algorithm to constrained SRA(c SRA) to carry out a further feature selection and lag optimization. During modeling stage, we introduce the Deep Neural Network(DNN) model in stock prediction under the suspicion that economic interactions are too complex for shallow networks to capture. Our experiments indeed show that deep structures generally perform better than shallow ones. Instead of comparing to a kitchen sink model, where over-fitting can easily happen with a shortage of data, we turn around and use a model ensemble approach which indirectly demonstrates our proposed method is adequate.展开更多
The tight-fractured gas reservoir of the Upper Triassic Xujiahe Formation in the Western Sichuan Depression has low porosity and permeability. This study presents a DNN-based method for identifying gas-bearing strata ...The tight-fractured gas reservoir of the Upper Triassic Xujiahe Formation in the Western Sichuan Depression has low porosity and permeability. This study presents a DNN-based method for identifying gas-bearing strata in tight sandstone. First, multi-component composite seismic attributes are obtained.The strong nonlinear relationships between multi-component composite attributes and gas-bearing reservoirs can be constrained through a DNN. Therefore, we identify and predict the gas-bearing strata using a DNN. Then, sample data are fed into the DNN for training and testing. After optimized network parameters are determined by the performance curves and empirical formulas, the best deep learning gas-bearing prediction model is determined. The composite seismic attributes can then be fed into the model to extrapolate the hydrocarbon-bearing characteristics from known drilling areas to the entire region for predicting the gas reservoir distribution. Finally, we assess the proposed method in terms of the structure and fracture characteristics and predict favorable exploration areas for identifying gas reservoirs.展开更多
Convolutional neural networks(CNNs)have shown tremendous progress and performance in recent years.Since emergence,CNNs have exhibited excellent performance in most of classification and segmentation tasks.Currently,th...Convolutional neural networks(CNNs)have shown tremendous progress and performance in recent years.Since emergence,CNNs have exhibited excellent performance in most of classification and segmentation tasks.Currently,the CNN family includes various architectures that dominate major vision-based recognition tasks.However,building a neural network(NN)by simply stacking convolution blocks inevitably limits its optimization ability and introduces overfitting and vanishing gradient problems.One of the key reasons for the aforementioned issues is network singularities,which have lately caused degenerating manifolds in the loss landscape.This situation leads to a slow learning process and lower performance.In this scenario,the skip connections turned out to be an essential unit of the CNN design to mitigate network singularities.The proposed idea of this research is to introduce skip connections in NN architecture to augment the information flow,mitigate singularities and improve performance.This research experimented with different levels of skip connections and proposed the placement strategy of these links for any CNN.To prove the proposed hypothesis,we designed an experimental CNN architecture,named as Shallow Wide ResNet or SRNet,as it uses wide residual network as a base network design.We have performed numerous experiments to assess the validity of the proposed idea.CIFAR-10 and CIFAR-100,two well-known datasets are used for training and testing CNNs.The final empirical results have shown a great many of promising outcomes in terms of performance,efficiency and reduction in network singularities issues.展开更多
Non-orthogonal multiple access(NOMA)has been a key enabling technology for the fifth generation(5G)cellular networks.Based on the NOMA principle,a traditional neural network has been implemented for user clustering(UC...Non-orthogonal multiple access(NOMA)has been a key enabling technology for the fifth generation(5G)cellular networks.Based on the NOMA principle,a traditional neural network has been implemented for user clustering(UC)to maximize the NOMA system’s throughput performance by considering that each sample is independent of the prior and the subsequent ones.Consequently,the prediction of UC for the future ones is based on the current clustering information,which is never used again due to the lack of memory of the network.Therefore,to relate the input features of NOMA users and capture the dependency in the clustering information,time-series methods can assist us in gaining a helpful insight into the future.Despite its mathematical complexity,the essence of time series comes down to examining past behavior and extending that information into the future.Hence,in this paper,we propose a novel and effective stacked long short term memory(S-LSTM)to predict the UC formation of NOMA users to enhance the throughput performance of the 5G-based NOMA systems.In the proposed strategy,the S-LSTM is modelled to handle the time-series input data to improve the predicting accuracy of UC of the NOMA users by implementing multiple LSTM layers with hidden cells.The implemented LSTM layers have feedback connections that help to capture the dependency in the clustering information as it propagates between the layers.Specifically,we develop,train,validate and test the proposed model to predict the UC formation for the futures ones by capturing the dependency in the clustering information based on the time-series data.Simulation results demonstrate that the proposed scheme effectively predicts UC and thereby attaining near-optimal throughput performance of 98.94%compared to the exhaustive search method.展开更多
Road safety performance function(SPF) analysis using data-driven and nonparametric methods, especially recent developed deep learning approaches, has gained increasing achievements. However, due to the learning mechan...Road safety performance function(SPF) analysis using data-driven and nonparametric methods, especially recent developed deep learning approaches, has gained increasing achievements. However, due to the learning mechanisms are hidden in a"black box" in deep learning, traffic features extraction and intelligent importance analysis are still unsolved and hard to generate.This paper focuses on this problem using a deciphered version of deep neural networks(DNN), one of the most popular deep learning models. This approach builds on visualization, feature importance and sensitivity analysis, can evaluate the contributions of input variables on model's "black box" feature learning process and output decision. Firstly, a visual feature importance(Vi FI) method that describes the importance of input features is proposed by adopting diagram and numerical-analysis. Secondly,by observing the change of weights using Vi FI on unsupervised training and fine-tuning of DNN, the final contributions of input features are calculated according to importance equations for both steps that we proposed. Sequentially, a case study based on a road SPF analysis is demonstrated, using data collected from a major Canadian highway, Highway 401. The proposed method allows effective deciphering of the model's inner workings and allows the significant features to be identified and the bad features to be eliminated. Finally, the revised dataset is used in crash modeling and vehicle collision prediction, and the testing result verifies that the deciphered and revised model achieves state-of-theart performance.展开更多
Because of computational complexity,the deep neural network(DNN)in embedded devices is usually trained on high-performance computers or graphic processing units(GPUs),and only the inference phase is implemented in emb...Because of computational complexity,the deep neural network(DNN)in embedded devices is usually trained on high-performance computers or graphic processing units(GPUs),and only the inference phase is implemented in embedded devices.Data processed by embedded devices,such as smartphones and wearables,are usually personalized,so the DNN model trained on public data sets may have poor accuracy when inferring the personalized data.As a result,retraining DNN with personalized data collected locally in embedded devices is necessary.Nevertheless,retraining needs labeled data sets,while the data collected locally are unlabeled,then how to retrain DNN with unlabeled data is a problem to be solved.This paper proves the necessity of retraining DNN model with personalized data collected in embedded devices after trained with public data sets.It also proposes a label generation method by which a fake label is generated for each unlabeled training case according to users’feedback,thus retraining can be performed with unlabeled data collected in embedded devices.The experimental results show that our fake label generation method has both good training effects and wide applicability.The advanced neural networks can be trained with unlabeled data from embedded devices and the individualized accuracy of the DNN model can be gradually improved along with personal using.展开更多
基金supported by the National Natural Science Foundation of China(Grant Nos.41772309 and 51908431)the Outstanding Youth Foundation of Hubei Province,China(Grant No.2019CFA074)。
文摘Real-time perception of rock mass information is of great importance to efficient tunneling and hazard prevention in tunnel boring machines(TBMs).In this study,a TBM-rock mutual feedback perception method based on data mining(DM) is proposed,which takes 10 tunneling parameters related to surrounding rock conditions as input features.For implementation,first,the database of TBM tunneling parameters was established,in which 10,807 tunneling cycles from the Songhua River water conveyance tunnel were accommodated.Then,the spectral clustering(SC) algorithm based on graph theory was introduced to cluster the TBM tunneling data.According to the clustering results and rock mass boreability index,the rock mass conditions were classified into four classes,and the reasonable distribution intervals of the main tunneling parameters corresponding to each class were presented.Meanwhile,based on the deep neural network(DNN),the real-time prediction model regarding different rock conditions was established.Finally,the rationality and adaptability of the proposed method were validated via analyzing the tunneling specific energy,feature importance,and training dataset size.The proposed TBM-rock mutual feedback perception method enables the automatic identification of rock mass conditions and the dynamic adjustment of tunneling parameters during TBM driving.Furthermore,in terms of the prediction performance,the method can predict the rock mass conditions ahead of the tunnel face in real time more accurately than the traditional machine learning prediction methods.
文摘In the past decade,dramatic progress has been made in the field of machine learning.This paper explores the possibility of applying deep learning in power system state estimation.Traditionally,physics-based models are used including weighted least square(WLS)or weighted least absolute value(WLAV).These models typically consider a single snapshot of the system without capturing temporal correlations of system states.In this paper,a physics-guided deep learning(PGDL)method is proposed.Specifically,inspired by autoencoders,deep neural networks(DNNs)are used to learn the temporal correlations.The estimated system states from DNNs are then checked against physics laws by running through a set of power flow equations.Hence,the proposed PGDL is both data-driven and physics-guided.The accuracy and robustness of the proposed PGDL method are compared with traditional methods in standard IEEE cases.Simulations show promising results and the applicability is further discussed.
文摘For training the present Neural Network(NN)models,the standard technique is to utilize decaying Learning Rates(LR).While the majority of these techniques commence with a large LR,they will decay multiple times over time.Decaying has been proved to enhance generalization as well as optimization.Other parameters,such as the network’s size,the number of hidden layers,drop-outs to avoid overfitting,batch size,and so on,are solely based on heuristics.This work has proposed Adaptive Teaching Learning Based(ATLB)Heuristic to identify the optimal hyperparameters for diverse networks.Here we consider three architec-tures Recurrent Neural Networks(RNN),Long Short Term Memory(LSTM),Bidirectional Long Short Term Memory(BiLSTM)of Deep Neural Networks for classification.The evaluation of the proposed ATLB is done through the various learning rate schedulers Cyclical Learning Rate(CLR),Hyperbolic Tangent Decay(HTD),and Toggle between Hyperbolic Tangent Decay and Triangular mode with Restarts(T-HTR)techniques.Experimental results have shown the performance improvement on the 20Newsgroup,Reuters Newswire and IMDB dataset.
基金supported by the NSF(No.DMS-1813071)(Chou)and the AFSOR(No.FA9550-22-1-0011)(Xiu).
文摘We present a numerical approach for modeling unknown dynamical systems using partially observed data,with a focus on biological systems with(relatively)complex dynamical behavior.As an extension of the recently developed deep neural network(DNN)learning methods,our approach is particularly suitable for practical situations when(i)measurement data are available for only a subset of the state variables,and(ii)the system parameters cannot be observed or measured at all.We demonstrate that,with a properly designed DNN structure with memory terms,effective DNN models can be learned from such partially observed data containing hidden parameters.The learned DNN model serves as an accurate predictive tool for system analysis.Through a few representative biological problems,we demonstrate that such DNN models can capture qualitative dynamical behavior changes in the system,such as bifurcations,even when the parameters controlling such behavior changes are completely unknown throughout not only the model learning process but also the system prediction process.The learned DNN model effectively creates a“closed”model involving only the observables when such a closed-form model does not exist mathematically.
基金Project supported by the National Natural Science Foundation of China(No.62376280)。
文摘Reinforcement learning(RL)has become a dominant decision-making paradigm and has achieved notable success in many real-world applications.Notably,deep neural networks play a crucial role in unlocking RL’s potential in large-scale decision-making tasks.Inspired by current major success of Transformer in natural language processing and computer vision,numerous bottlenecks have been overcome by combining Transformer with RL for decision-making.This paper presents a multiangle systematic survey of various Transformer-based RL(TransRL)models applied in decision-making tasks,including basic models,advanced algorithms,representative implementation instances,typical applications,and known challenges.Our work aims to provide insights into problems that inherently arise with the current RL approaches,and examines how we can address them with better TransRL models.To our knowledge,we are the first to present a comprehensive review of the recent Transformer research developments in RL for decision-making.We hope that this survey provides a comprehensive review of TransRL models and inspires the RL community in its pursuit of future directions.To keep track of the rapid TransRL developments in the decision-making domains,we summarize the latest papers and their open-source implementations at https://github.com/williamyuanv0/Transformer-in-Reinforcement-Learning-for-Decision-Making-A-Survey.
基金supported by Aviation Industry Corporation of China(AVIC)Manufacturing Technology Institute(MTI)and in part by China Scholarship Council(CSC)(201908060236)。
文摘Sheet metal forming technologies have been intensively studied for decades to meet the increasing demand for lightweight metal components.To surmount the springback occurring in sheet metal forming processes,numerous studies have been performed to develop compensation methods.However,for most existing methods,the development cycle is still considerably time-consumptive and demands high computational or capital cost.In this paper,a novel theory-guided regularization method for training of deep neural networks(DNNs),implanted in a learning system,is introduced to learn the intrinsic relationship between the workpiece shape after springback and the required process parameter,e.g.,loading stroke,in sheet metal bending processes.By directly bridging the workpiece shape to the process parameter,issues concerning springback in the process design would be circumvented.The novel regularization method utilizes the well-recognized theories in material mechanics,Swift’s law,by penalizing divergence from this law throughout the network training process.The regularization is implemented by a multi-task learning network architecture,with the learning of extra tasks regularized during training.The stress-strain curve describing the material properties and the prior knowledge used to guide learning are stored in the database and the knowledge base,respectively.One can obtain the predicted loading stroke for a new workpiece shape by importing the target geometry through the user interface.In this research,the neural models were found to outperform a traditional machine learning model,support vector regression model,in experiments with different amount of training data.Through a series of studies with varying conditions of training data structure and amount,workpiece material and applied bending processes,the theory-guided DNN has been shown to achieve superior generalization and learning consistency than the data-driven DNNs,especially when only scarce and scattered experiment data are available for training which is often the c
基金supported by National Natural Science Foundation of China(No.61873142)the Science and Technology Research Program of the Chongqing Municipal Education Commission,China(Nos.KJZD-K202201901,KJQN202201109,KJQN202101904,KJQN202001903 and CXQT21035)+2 种基金the Scientific Research Foundation of Chongqing University of Technology,China(No.2019ZD76)the Scientific Research Foundation of Chongqing Institute of Engineering,China(No.2020xzky05)the Chongqing Municipal Natural Science Foundation,China(No.cstc2020jcyj-msxmX0666).
文摘In industrial process control systems,there is overwhelming evidence corroborating the notion that economic or technical limitations result in some key variables that are very difficult to measure online.The data-driven soft sensor is an effective solution because it provides a reliable and stable online estimation of such variables.This paper employs a deep neural network with multiscale feature extraction layers to build soft sensors,which are applied to the benchmarked Tennessee-Eastman process(TEP)and a real wind farm case.The comparison of modelling results demonstrates that the multiscale feature extraction layers have the following advantages over other methods.First,the multiscale feature extraction layers significantly reduce the number of parameters compared to the other deep neural networks.Second,the multiscale feature extraction layers can powerfully extract dataset characteristics.Finally,the multiscale feature extraction layers with fully considered historical measurements can contain richer useful information and improved representation compared to traditional data-driven models.
基金National Natural Science Foundation of China(Nos.61931011,62072303 and 61872310)the Key-area Research and Development Program of Guangdong Province,China(No.2021B0101400003)+2 种基金Hong Kong Research Grants Council(RGC)Research Impact Fund,China(No.R5060-19)General Research Fund(Nos.152221/19E,152203/20E and 152244/2IE)Shenzhen Science and Technology Innovation Commission,China(No.JCYJ20200109142008673).
文摘With the vigorous development of artificial intelligence(AI),intelligence applications based on deep neural networks(DNNs)have changed people’s lifestyles and production efficiency.However,the large amount of computation and data generated from the network edge becomes the major bottleneck,and the traditional cloud-based computing mode has been unable to meet the requirements of realtime processing tasks.To solve the above problems,by embedding AI model training and inference capabilities into the network edge,edge intelligence(EI)becomes a cutting-edge direction in the field of AI.Furthermore,collaborative DNN inference among the cloud,edge,and end devices provides a promising way to boost EI.Nevertheless,at present,EI oriented collaborative DNN inference is still in its early stage,lacking systematic classification and discussion of existing research efforts.Motivated by it,we have comprehensively investigated recent studies on EI-oriented collaborative DNN inference.In this paper,we first review the background and motivation of EI.Then,we classify four typical collaborative DNN inference paradigms for EI,and analyse their characteristics and key technologies.Finally,we summarize the current challenges of collaborative DNN inference,discuss future development trends and provide future research directions.
基金supported in part by Science and Technology Projects of Electric Power Research Institute of State Grid Jiangsu Electric Power Co.,Ltd.(J2021171).
文摘The access of unified power flow controllers(UPFC)has changed the structure and operation mode of power grids all across the world,and it has brought severe challenges to the traditional real-time calculation of security correction based on traditionalmodels.Considering the limitation of computational efficiency regarding complex,physical models,a data-driven power system security correction method with UPFC is,in this paper,proposed.Based on the complex mapping relationship between the operation state data and the security correction strategy,a two-stage deep neural network(DNN)learning framework is proposed,which divides the offline training task of security correction into two stages:in the first stage,the stacked auto-encoder(SAE)classification model is established,and the node correction state(0/1)output based on the fault information;in the second stage,the DNN learningmodel is established,and the correction amount of each action node is obtained based on the action nodes output in the previous stage.In this paper,the UPFC demonstration project of NanjingWest Ring Network is taken as a case study to validate the proposed method.The results show that the proposed method can fully meet the real-time security correction time requirements of power grids,and avoid the inherent defects of the traditional model method without an iterative solution and can also provide reasonable security correction strategies for N-1 and N-2 faults.
基金supported by the National Natural Science Foundation of China(Grant Nos.11822205 and 11772167).
文摘Real-time guidance is critical for the vertical recovery of rockets.However,traditional sequential convex optimization algorithms suffer from shortcomings in terms of their poor real-time performance.This work focuses on applying the deep learning-based closedloop guidance algorithm and error propagation analysis for powered landing,thereby significantly improving the real-time performance.First,a controller consisting of two deep neural networks is constructed to map the thrust direction and magnitude of the rocket according to the state variables.Thereafter,the analytical transition relationships between different uncertainty sources and the state propagation error in a single guidance period are analyzed by adopting linear covariance analysis.Finally,the accuracy of the proposed methods is verified via a comparison with the indirect method and Monte Carlo simulations.Compared with the traditional sequential convex optimization algorithm,our method reduces the computation time from 75 ms to less than 1 ms.Therefore,it shows potential for online applications.
基金supported by the National Defense Science and Technology Innovation(18-163-15-LZ-001-004-13).
文摘This paper investigates the guidance method based on reinforcement learning(RL)for the coplanar orbital interception in a continuous low-thrust scenario.The problem is formulated into a Markov decision process(MDP)model,then a welldesigned RL algorithm,experience based deep deterministic policy gradient(EBDDPG),is proposed to solve it.By taking the advantage of prior information generated through the optimal control model,the proposed algorithm not only resolves the convergence problem of the common RL algorithm,but also successfully trains an efficient deep neural network(DNN)controller for the chaser spacecraft to generate the control sequence.Numerical simulation results show that the proposed algorithm is feasible and the trained DNN controller significantly improves the efficiency over traditional optimization methods by roughly two orders of magnitude.
基金supported by Shanghai Science and Technology Commission(21ZR1442400)the National Natural Science Foundation of China(31771373)startup funding from ShanghaiTech University.
文摘DNA methylation analysis has been applied to determine the primary site of cancer;however, robust and accurate prediction of cancer types with a minimum number of sites is still a significant scientific challenge. To build an accurate and robust cancer type prediction tool with a minimum number of DNA methylation sites, we internally benchmarked different DNA methylation site selection and ranking procedures, as well as different classification models. We used The Cancer Genome Atlas dataset (26 cancer types with 8296 samples) to train and test models and used an independent dataset (17 cancer types with 2738 samples) for model validation. A deep neural network model using a combined feature selection procedure (named MethyDeep) can predict 26 cancer types using 30 methylation sites with superior performance compared with the known methods for both primary and metastatic cancers in independent validation datasets. In conclusion, MethyDeep is an accurate and robust cancer type predictor with the minimum number of DNA methylation sites;it could help the cost-effective clarification of cancer of unknown primary patients and the liquid biopsy-based early screening of cancers.
文摘Distributed denial of service(DDoS)attack is the most common attack that obstructs a network and makes it unavailable for a legitimate user.We proposed a deep neural network(DNN)model for the detection of DDoS attacks in the Software-Defined Networking(SDN)paradigm.SDN centralizes the control plane and separates it from the data plane.It simplifies a network and eliminates vendor specification of a device.Because of this open nature and centralized control,SDN can easily become a victim of DDoS attacks.We proposed a supervised Developed Deep Neural Network(DDNN)model that can classify the DDoS attack traffic and legitimate traffic.Our Developed Deep Neural Network(DDNN)model takes a large number of feature values as compared to previously proposed Machine Learning(ML)models.The proposed DNN model scans the data to find the correlated features and delivers high-quality results.The model enhances the security of SDN and has better accuracy as compared to previously proposed models.We choose the latest state-of-the-art dataset which consists of many novel attacks and overcomes all the shortcomings and limitations of the existing datasets.Our model results in a high accuracy rate of 99.76%with a low false-positive rate and 0.065%low loss rate.The accuracy increases to 99.80%as we increase the number of epochs to 100 rounds.Our proposed model classifies anomalous and normal traffic more accurately as compared to the previously proposed models.It can handle a huge amount of structured and unstructured data and can easily solve complex problems.
基金supported in part by National Science Fund under Grant No.61273365111 Project under Grant No.B08004
文摘Video Description aims to automatically generate descriptive natural language for videos.Due to the large volume of multi-modal data and successful implementations of Deep Neural Networks(DNNs),a wide range of models have been proposed.However,previous models learn insufficient linguistic information or correlation between visual and textual modalities.In order to address those problems,this paper proposes an integrated model using Long Short-Term Memory(LSTM).This proposed model consists of triple channels in parallel:a primary video description channel,a sentence-to-sentence channel for language learning,and a channel to integrate visual and textual information.Additionally,the parallel three channels are connected by LSTM weight matrices during training.The VD-ivt model is evaluated on two publicly available datasets,i.e.Youtube2 Text and LSMDC.Experimental results demonstrate that the performance of the proposed model outperforms those benchmarks.
基金the National Natural Science Foundation of China(Nos.11501355 and 71571116)the Project of Knowledge Innovation Program of Shanghai Municipal Education Commission(No.15ZZ090)+2 种基金the 59th China Postdoctoral Sciences Foundation Funded Project(No.2016M591640)the Humanities and Social Sciences Research Project of Ministry of Education(No.15YJA790039)the National Social Science Foundation of China(No.15ZDA058)
文摘The literature generally agrees that longer-horizon(over a month) predictions make more sense than short-horizon ones. However, it's an especially challenging task due to the lack of data(in unit of long horizon)and economic data have a low S/N ratio. We hypothesize that the stock trend is largely dictated by driving factors which are filtered by psychological factors and work on behavioral factors: representative indicators from these three aspects would be adequate in trend prediction. We then extend the Stepwise Regression Analysis(SRA)algorithm to constrained SRA(c SRA) to carry out a further feature selection and lag optimization. During modeling stage, we introduce the Deep Neural Network(DNN) model in stock prediction under the suspicion that economic interactions are too complex for shallow networks to capture. Our experiments indeed show that deep structures generally perform better than shallow ones. Instead of comparing to a kitchen sink model, where over-fitting can easily happen with a shortage of data, we turn around and use a model ensemble approach which indirectly demonstrates our proposed method is adequate.
基金funded by the Natural Science Foundation of Shandong Province (ZR202103050722)National Natural Science Foundation of China (41174098)。
文摘The tight-fractured gas reservoir of the Upper Triassic Xujiahe Formation in the Western Sichuan Depression has low porosity and permeability. This study presents a DNN-based method for identifying gas-bearing strata in tight sandstone. First, multi-component composite seismic attributes are obtained.The strong nonlinear relationships between multi-component composite attributes and gas-bearing reservoirs can be constrained through a DNN. Therefore, we identify and predict the gas-bearing strata using a DNN. Then, sample data are fed into the DNN for training and testing. After optimized network parameters are determined by the performance curves and empirical formulas, the best deep learning gas-bearing prediction model is determined. The composite seismic attributes can then be fed into the model to extrapolate the hydrocarbon-bearing characteristics from known drilling areas to the entire region for predicting the gas reservoir distribution. Finally, we assess the proposed method in terms of the structure and fracture characteristics and predict favorable exploration areas for identifying gas reservoirs.
文摘Convolutional neural networks(CNNs)have shown tremendous progress and performance in recent years.Since emergence,CNNs have exhibited excellent performance in most of classification and segmentation tasks.Currently,the CNN family includes various architectures that dominate major vision-based recognition tasks.However,building a neural network(NN)by simply stacking convolution blocks inevitably limits its optimization ability and introduces overfitting and vanishing gradient problems.One of the key reasons for the aforementioned issues is network singularities,which have lately caused degenerating manifolds in the loss landscape.This situation leads to a slow learning process and lower performance.In this scenario,the skip connections turned out to be an essential unit of the CNN design to mitigate network singularities.The proposed idea of this research is to introduce skip connections in NN architecture to augment the information flow,mitigate singularities and improve performance.This research experimented with different levels of skip connections and proposed the placement strategy of these links for any CNN.To prove the proposed hypothesis,we designed an experimental CNN architecture,named as Shallow Wide ResNet or SRNet,as it uses wide residual network as a base network design.We have performed numerous experiments to assess the validity of the proposed idea.CIFAR-10 and CIFAR-100,two well-known datasets are used for training and testing CNNs.The final empirical results have shown a great many of promising outcomes in terms of performance,efficiency and reduction in network singularities issues.
基金This work was funded by Multimedia University under Grant Number MMUI/170084.
文摘Non-orthogonal multiple access(NOMA)has been a key enabling technology for the fifth generation(5G)cellular networks.Based on the NOMA principle,a traditional neural network has been implemented for user clustering(UC)to maximize the NOMA system’s throughput performance by considering that each sample is independent of the prior and the subsequent ones.Consequently,the prediction of UC for the future ones is based on the current clustering information,which is never used again due to the lack of memory of the network.Therefore,to relate the input features of NOMA users and capture the dependency in the clustering information,time-series methods can assist us in gaining a helpful insight into the future.Despite its mathematical complexity,the essence of time series comes down to examining past behavior and extending that information into the future.Hence,in this paper,we propose a novel and effective stacked long short term memory(S-LSTM)to predict the UC formation of NOMA users to enhance the throughput performance of the 5G-based NOMA systems.In the proposed strategy,the S-LSTM is modelled to handle the time-series input data to improve the predicting accuracy of UC of the NOMA users by implementing multiple LSTM layers with hidden cells.The implemented LSTM layers have feedback connections that help to capture the dependency in the clustering information as it propagates between the layers.Specifically,we develop,train,validate and test the proposed model to predict the UC formation for the futures ones by capturing the dependency in the clustering information based on the time-series data.Simulation results demonstrate that the proposed scheme effectively predicts UC and thereby attaining near-optimal throughput performance of 98.94%compared to the exhaustive search method.
基金supported by the National Science and Engineering Research Council of Canada(NSERC)Ontario Research Fund–Research Excellence(ORF-RE)+3 种基金the Ministry of Transportation Ontario(MTO)through Its Highway Infrastructure Innovation Funding Program(HIIFP)Beijing Postdoctoral Science Foundation(ZZ-2019-65)Beijing Chaoyang District Postdoctoral Science Foundation(2019ZZ-45)Beijing Municipal Education Commission(KM201811232016)。
文摘Road safety performance function(SPF) analysis using data-driven and nonparametric methods, especially recent developed deep learning approaches, has gained increasing achievements. However, due to the learning mechanisms are hidden in a"black box" in deep learning, traffic features extraction and intelligent importance analysis are still unsolved and hard to generate.This paper focuses on this problem using a deciphered version of deep neural networks(DNN), one of the most popular deep learning models. This approach builds on visualization, feature importance and sensitivity analysis, can evaluate the contributions of input variables on model's "black box" feature learning process and output decision. Firstly, a visual feature importance(Vi FI) method that describes the importance of input features is proposed by adopting diagram and numerical-analysis. Secondly,by observing the change of weights using Vi FI on unsupervised training and fine-tuning of DNN, the final contributions of input features are calculated according to importance equations for both steps that we proposed. Sequentially, a case study based on a road SPF analysis is demonstrated, using data collected from a major Canadian highway, Highway 401. The proposed method allows effective deciphering of the model's inner workings and allows the significant features to be identified and the bad features to be eliminated. Finally, the revised dataset is used in crash modeling and vehicle collision prediction, and the testing result verifies that the deciphered and revised model achieves state-of-theart performance.
基金supported by the National Natural Science Foundation of China under Grants No.61534002,No.61761136015,No.61701095.
文摘Because of computational complexity,the deep neural network(DNN)in embedded devices is usually trained on high-performance computers or graphic processing units(GPUs),and only the inference phase is implemented in embedded devices.Data processed by embedded devices,such as smartphones and wearables,are usually personalized,so the DNN model trained on public data sets may have poor accuracy when inferring the personalized data.As a result,retraining DNN with personalized data collected locally in embedded devices is necessary.Nevertheless,retraining needs labeled data sets,while the data collected locally are unlabeled,then how to retrain DNN with unlabeled data is a problem to be solved.This paper proves the necessity of retraining DNN model with personalized data collected in embedded devices after trained with public data sets.It also proposes a label generation method by which a fake label is generated for each unlabeled training case according to users’feedback,thus retraining can be performed with unlabeled data collected in embedded devices.The experimental results show that our fake label generation method has both good training effects and wide applicability.The advanced neural networks can be trained with unlabeled data from embedded devices and the individualized accuracy of the DNN model can be gradually improved along with personal using.
