ST-LSTM-SA:A New Ocean Sound Velocity Field Prediction Model Based on Deep Learning

The scarcity of in-situ ocean observations poses a challenge for real-time information acquisition in the ocean.Among the crucial hydroacoustic environmental parameters,ocean sound velocity exhibits significant spatial and temporal variability and it is highly relevant to oceanic research.In this study,we propose a new data-driven approach,leveraging deep learning techniques,for the prediction of sound velocity fields(SVFs).Our novel spatiotemporal prediction model,STLSTM-SA,combines Spatiotemporal Long Short-Term Memory(ST-LSTM) with a self-attention mechanism to enable accurate and real-time prediction of SVFs.To circumvent the limited amount of observational data,we employ transfer learning by first training the model using reanalysis datasets,followed by fine-tuning it using in-situ analysis data to obtain the final prediction model.By utilizing the historical 12-month SVFs as input,our model predicts the SVFs for the subsequent three months.We compare the performance of five models:Artificial Neural Networks(ANN),Long ShortTerm Memory(LSTM),Convolutional LSTM(ConvLSTM),ST-LSTM,and our proposed ST-LSTM-SA model in a test experiment spanning 2019 to 2022.Our results demonstrate that the ST-LSTM-SA model significantly improves the prediction accuracy and stability of sound velocity in both temporal and spatial dimensions.The ST-LSTM-SA model not only accurately predicts the ocean sound velocity field(SVF),but also provides valuable insights for spatiotemporal prediction of other oceanic environmental variables.

Battery pack capacity estimation for electric vehicles based on enhanced machine learning and field data

Accurate capacity estimation is of great importance for the reliable state monitoring,timely maintenance,and second-life utilization of lithium-ion batteries.Despite numerous works on battery capacity estimation using laboratory datasets,most of them are applied to battery cells and lack satisfactory fidelity when extended to real-world electric vehicle(EV)battery packs.The challenges intensify for large-sized EV battery packs,where unpredictable operating profiles and low-quality data acquisition hinder precise capacity estimation.To fill the gap,this study introduces a novel data-driven battery pack capacity estimation method grounded in field data.The proposed approach begins by determining labeled capacity through an innovative combination of the inverse ampere-hour integral,open circuit voltage-based,and resistance-based correction methods.Then,multiple health features are extracted from incremental capacity curves,voltage curves,equivalent circuit model parameters,and operating temperature to thoroughly characterize battery aging behavior.A feature selection procedure is performed to determine the optimal feature set based on the Pearson correlation coefficient.Moreover,a convolutional neural network and bidirectional gated recurrent unit,enhanced by an attention mechanism,are employed to estimate the battery pack capacity in real-world EV applications.Finally,the proposed method is validated with a field dataset from two EVs,covering approximately 35,000 kilometers.The results demonstrate that the proposed method exhibits better estimation performance with an error of less than 1.1%compared to existing methods.This work shows great potential for accurate large-sized EV battery pack capacity estimation based on field data,which provides significant insights into reliable labeled capacity calculation,effective features extraction,and machine learning-enabled health diagnosis.

Reconstruction of poloidal magnetic field profiles in field-reversed configurations with machine learning in laser-driven ion-beam trace probe

The diagnostic of poloidal magnetic field(B_(p))in field-reversed configuration(FRC),promising for achieving efficient plasma confinement due to its highβ,is a huge challenge because B_(p)is small and reverses around the core region.The laser-driven ion-beam trace probe(LITP)has been proven to diagnose the B_(p)profile in FRCs recently,whereas the existing iterative reconstruction approach cannot handle the measurement errors well.In this work,the machine learning approach,a fast-growing and powerful technology in automation and control,is applied to B_(p)reconstruction in FRCs based on LITP principles and it has a better performance than the previous approach.The machine learning approach achieves a more accurate reconstruction of B_(p)profile when 20%detector errors are considered,15%B_(p)fluctuation is introduced and the size of the detector is remarkably reduced.Therefore,machine learning could be a powerful support for LITP diagnosis of the magnetic field in magnetic confinement fusion devices.

A Deep Learning Method to Process Scattered Field Data in Biomedical Imaging System

This paper proposed a deep-learning-based method to process the scattered field data of transmitting antenna,which is unmeasurable in inverse scattering system because the transmitting and receiving antennas are multiplexed.A U-net convolutional neural network(CNN)is used to recover the scattered field data of each transmitting antenna.The numerical results proved that the proposed method can complete the scattered field data at the transmitting antenna which is unable to measure in the actual experiment and can also eliminate the reconstructed error caused by the loss of scattered field data.

Retinal vascular morphological characteristics in diabetic retinopathy: an artificial intelligence study using a transfer learning system to analyze ultra-wide field images

AIM:To investigate the morphological characteristics of retinal vessels in patients with different severity of diabetic retinopathy(DR)and in patients with or without diabetic macular edema(DME).METHODS:The 239 eyes of DR patients and 100 eyes of healthy individuals were recruited for the study.The severity of DR patients was graded as mild,moderate and severe non-proliferative diabetic retinopathy(NPDR)according to the international clinical diabetic retinopathy(ICDR)disease severity scale classification,and retinal vascular morphology was quantitatively analyzed in ultra-wide field images using RU-net and transfer learning methods.The presence of DME was determined by optical coherence tomography(OCT),and differences in vascular morphological characteristics were compared between patients with and without DME.RESULTS:Retinal vessel segmentation using RU-net and transfer learning system had an accuracy of 99%and a Dice metric of 0.76.Compared with the healthy group,the DR group had smaller vessel angles(33.68±3.01 vs 37.78±1.60),smaller fractal dimension(Df)values(1.33±0.05 vs 1.41±0.03),less vessel density(1.12±0.44 vs 2.09±0.36)and fewer vascular branches(206.1±88.8 vs 396.5±91.3),all P<0.001.As the severity of DR increased,Df values decreased,P=0.031.No significant difference between the DME and non-DME groups were observed in vascular morphological characteristics.CONCLUSION:In this study,an artificial intelligence retinal vessel segmentation system is used with 99%accuracy,thus providing with relatively satisfactory performance in the evaluation of quantitative vascular morphology.DR patients have a tendency of vascular occlusion and dropout.The presence of DME does not compromise the integral retinal vascular pattern.

Curriculum Reform of Programming and Algorithm Foundation for Competency-based Learning

It is important to transform knowledge-based learning to competency-based learning.This paper describes the exploration and practice of“programming and algorithm foundation”curriculum reform for competency-based learning.In order to cultivate students’ability of high-level program development,the intelligent learning system of“MOOC/SPOC+icoding online experiment and programming ability test Platform+Rain Classroom”is established.In the case of limited class hours,we make full use of online resources to build a student-centered method to internalize knowledge and ability.We guide students to complete the basic knowledge module of MOOC or SPOC,and complete the programming experiment on icoding platform.According to the feedback of learning outcome,teachers use offline classroom and rain classroom to sort out the key and difficult points,expanding the depth and breadth of the curriculum,and stimulate students’enthusiasm to participate in the curriculum.

Machine learning molecular dynamics simulations of liquid methanol

As the simplest hydrogen-bonded alcohol,liquid methanol has attracted intensive experimental and theoretical interest.However,theoretical investigations on this system have primarily relied on empirical intermolecular force fields or ab initio molecular dynamics with semilocal density functionals.Inspired by recent studies on bulk water using increasingly accurate machine learning force fields,we report a new machine learning force field for liquid methanol with a hybrid functional revPBE0 plus dispersion correction.Molecular dynamics simulations on this machine learning force field are orders of magnitude faster than ab initio molecular dynamics simulations,yielding the radial distribution functions,selfdiffusion coefficients,and hydrogen bond network properties with very small statistical errors.The resulting structural and dynamical properties are compared well with the experimental data,demonstrating the superior accuracy of this machine learning force field.This work represents a successful step toward a first-principles description of this benchmark system and showcases the general applicability of the machine learning force field in studying liquid systems.

Solution to reinforcement learning problems with artificial potential field

A novel method was designed to solve reinforcement learning problems with artificial potential field.Firstly a reinforcement learning problem was transferred to a path planning problem by using artificial potential field(APF),which was a very appropriate method to model a reinforcement learning problem.Secondly,a new APF algorithm was proposed to overcome the local minimum problem in the potential field methods with a virtual water-flow concept.The performance of this new method was tested by a gridworld problem named as key and door maze.The experimental results show that within 45 trials,good and deterministic policies are found in almost all simulations.In comparison with WIERING's HQ-learning system which needs 20 000 trials for stable solution,the proposed new method can obtain optimal and stable policy far more quickly than HQ-learning.Therefore,the new method is simple and effective to give an optimal solution to the reinforcement learning problem.

Uncertainty-Aware Physical Simulation of Neural Radiance Fields for Fluids

This paper presents a novel framework aimed at quantifying uncertainties associated with the 3D reconstruction of smoke from2Dimages.This approach reconstructs color and density fields from 2D images using Neural Radiance Field(NeRF)and improves image quality using frequency regularization.The NeRF model is obtained via joint training ofmultiple artificial neural networks,whereby the expectation and standard deviation of density fields and RGB values can be evaluated for each pixel.In addition,customized physics-informed neural network(PINN)with residual blocks and two-layer activation functions are utilized to input the density fields of the NeRF into Navier-Stokes equations and convection-diffusion equations to reconstruct the velocity field.The velocity uncertainties are also evaluated through ensemble learning.The effectiveness of the proposed algorithm is demonstrated through numerical examples.The presentmethod is an important step towards downstream tasks such as reliability analysis and robust optimization in engineering design.

MUS Model:A Deep Learning-Based Architecture for IoT Intrusion Detection

In the face of the effective popularity of the Internet of Things(IoT),but the frequent occurrence of cybersecurity incidents,various cybersecurity protection means have been proposed and applied.Among them,Intrusion Detection System(IDS)has been proven to be stable and efficient.However,traditional intrusion detection methods have shortcomings such as lowdetection accuracy and inability to effectively identifymalicious attacks.To address the above problems,this paper fully considers the superiority of deep learning models in processing highdimensional data,and reasonable data type conversion methods can extract deep features and detect classification using advanced computer vision techniques to improve classification accuracy.TheMarkov TransformField(MTF)method is used to convert 1Dnetwork traffic data into 2D images,and then the converted 2D images are filtered by UnsharpMasking to enhance the image details by sharpening;to further improve the accuracy of data classification and detection,unlike using the existing high-performance baseline image classification models,a soft-voting integrated model,which integrates three deep learning models,MobileNet,VGGNet and ResNet,to finally obtain an effective IoT intrusion detection architecture:the MUS model.Four types of experiments are conducted on the publicly available intrusion detection dataset CICIDS2018 and the IoT network traffic dataset N_BaIoT,and the results demonstrate that the accuracy of attack traffic detection is greatly improved,which is not only applicable to the IoT intrusion detection environment,but also to different types of attacks and different network environments,which confirms the effectiveness of the work done.

Deep-learning classifier with ultrawide-field fundus ophthalmoscopy for detecting branch retinal vein occlusion

AIM: To investigate and compare the efficacy of two machine-learning technologies with deep-learning(DL) and support vector machine(SVM) for the detection of branch retinal vein occlusion(BRVO) using ultrawide-field fundus images. METHODS: This study included 237 images from 236 patients with BRVO with a mean±standard deviation of age 66.3±10.6 y and 229 images from 176 non-BRVO healthy subjects with a mean age of 64.9±9.4 y. Training was conducted using a deep convolutional neural network using ultrawide-field fundus images to construct the DL model. The sensitivity, specificity, positive predictive value(PPV), negative predictive value(NPV) and area under the curve(AUC) were calculated to compare the diagnostic abilities of the DL and SVM models. RESULTS: For the DL model, the sensitivity, specificity, PPV, NPV and AUC for diagnosing BRVO was 94.0%(95%CI: 93.8%-98.8%), 97.0%(95%CI: 89.7%-96.4%), 96.5%(95%CI: 94.3%-98.7%), 93.2%(95%CI: 90.5%-96.0%) and 0.976(95%CI: 0.960-0.993), respectively. In contrast, for the SVM model, these values were 80.5%(95%CI: 77.8%-87.9%), 84.3%(95%CI: 75.8%-86.1%), 83.5%(95%CI: 78.4%-88.6%), 75.2%(95%CI: 72.1%-78.3%) and 0.857(95%CI: 0.811-0.903), respectively. The DL model outperformed the SVM model in all the aforementioned parameters(P<0.001). CONCLUSION: These results indicate that the combination of the DL model and ultrawide-field fundus ophthalmoscopy may distinguish between healthy and BRVO eyes with a high level of accuracy. The proposed combination may be used for automatically diagnosing BRVO in patients residing in remote areas lacking access to an ophthalmic medical center.

基于改进Q-learning算法的移动机器人路径规划

针对传统Q-learning算法应用在路径规划中存在收敛速度慢、运行时间长、学习效率差等问题,提出一种将人工势场法和传统Q-learning算法结合的改进Q-learning算法。该算法引入人工势场法的引力函数与斥力函数,通过对比引力函数动态选择奖励值,以及对比斥力函数计算姿值,动态更新Q值,使移动机器人具有目的性的探索,并且优先选择离障碍物较远的位置移动。通过仿真实验证明,与传统Q-learning算法、引入引力场算法对比,改进Q-learning算法加快了收敛速度,缩短了运行时间,提高了学习效率,降低了与障碍物相撞的概率,使移动机器人能够快速地找到一条无碰撞通路。

Enhancing economic sustainability in mature oil fields:Insights from the clustering approach to select candidate wells for extended shut-in

Fluctuations in oil prices adversely affect decision making situations in which performance forecasting must be combined with realistic price forecasts.In periods of significant price drops,companies may consider extended duration of well shut-ins(i.e.temporarily stopping oil production)for economic reasons.For example,prices during the early days of the Covid-19 pandemic forced operators to consider shutting in all or some of their active wells.In the case of partial shut-in,selection of candidate wells may evolve as a challenging decision problem considering the uncertainties involved.In this study,a mature oil field with a long(50+years)production history with 170+wells is considered.Reservoirs with similar conditions face many challenges related to economic sustainability such as frequent maintenance requirements and low production rates.We aimed to solve this decision-making problem through unsupervised machine learning.Average reservoir characteristics at well locations,well production performance statistics and well locations are used as potential features that could characterize similarities and differences among wells.While reservoir characteristics are measured at well locations for the purpose of describing the subsurface reservoir,well performance consists of volumetric rates and pressures,which are frequently measured during oil production.After a multivariate data analysis that explored correlations among parameters,clustering algorithms were used to identify groups of wells that are similar with respect to aforementioned features.Using the field’s reservoir simulation model,scenarios of shutting in different groups of wells were simulated.Forecasted reservoir performance for three years was used for economic evaluation that assumed an oil price drop to$30/bbl for 6,12 or 18 months.Results of economic analysis were analyzed to identify which group(s)of wells should have been shut-in by also considering the sensitivity to different price levels.It was observed that wells can be characterized in the 3-cluster case as low,medium and high performance wells.Analyzing the forecasting scenarios showed that shutting in all or high-and medium-performance wells altogether results in better economic outcomes.The results were most sensitive to the number of active wells and the oil price during the high-price period.This study demonstrated the effectiveness of unsupervised machine learning in well classification for operational decision making purposes.Operating companies may use this approach for improved decision making to select wells for extended shut-in during low oil-price periods.This approach would lead to cost savings especially in mature fields with low-profit margins.

Multi-step Reinforcement Learning Algorithm of Mobile Robot Path Planning Based on Virtual Potential Field

A algorithm of dynamic multi-step reinforcement learning based on virtual potential field path planning is proposed in this paper. Firstly, it is constructed the virtual potential field according to the known information. And then in view of Q learning algorithm of the QekT algorithm, a multi-step reinforcement learning algorithm is proposed in this paper. It can update current Q value used of future dynamic k steps according to the current environment status. At the same time, the convergence is analyzed. Finally the simulation experiments are done. It shows that the proposed algorithm and convergence and so on are more efficiency than similar algorithms.

Machine learning inspired workflow to revise field development plan under uncertainty

We present an efficient and risk-informed closed-loop field development (CLFD) workflow for recurrently revising the field development plan (FDP) using the accrued information. To make the process practical, we integrated multiple concepts of machine learning, an intelligent selection process to discard the worst FDP options and a growing set of representative reservoir models. These concepts were combined and used with a cluster-based learning and evolution optimizer to efficiently explore the search space of decision variables. Unlike previous studies, we also added the execution time of the CLFD workflow and worked with more realistic timelines to confirm the utility of a CLFD workflow. To appreciate the importance of data assimilation and new well-logs in a CLFD workflow, we carried out researches at rigorous conditions without a reduction in uncertainty attributes. The proposed CLFD workflow was implemented on a benchmark analogous to a giant field with extensively time-consuming simulation models. The results underscore that an ensemble with as few as 100 scenarios was sufficient to gauge the geological uncertainty, despite working with a giant field with highly heterogeneous characteristics. It is demonstrated that the CLFD workflow can improve the efficiency by over 85% compared to the previously validated workflow. Finally, we present some acute insights and problems related to data assimilation for the practical application of a CLFD workflow.

Magnetic Field-Based Reward Shaping for Goal-Conditioned Reinforcement Learning

Goal-conditioned reinforcement learning(RL)is an interesting extension of the traditional RL framework,where the dynamic environment and reward sparsity can cause conventional learning algorithms to fail.Reward shaping is a practical approach to improving sample efficiency by embedding human domain knowledge into the learning process.Existing reward shaping methods for goal-conditioned RL are typically built on distance metrics with a linear and isotropic distribution,which may fail to provide sufficient information about the ever-changing environment with high complexity.This paper proposes a novel magnetic field-based reward shaping(MFRS)method for goal-conditioned RL tasks with dynamic target and obstacles.Inspired by the physical properties of magnets,we consider the target and obstacles as permanent magnets and establish the reward function according to the intensity values of the magnetic field generated by these magnets.The nonlinear and anisotropic distribution of the magnetic field intensity can provide more accessible and conducive information about the optimization landscape,thus introducing a more sophisticated magnetic reward compared to the distance-based setting.Further,we transform our magnetic reward to the form of potential-based reward shaping by learning a secondary potential function concurrently to ensure the optimal policy invariance of our method.Experiments results in both simulated and real-world robotic manipulation tasks demonstrate that MFRS outperforms relevant existing methods and effectively improves the sample efficiency of RL algorithms in goal-conditioned tasks with various dynamics of the target and obstacles.

Transfer-Learning for Automated Seizure Detection Based on Electric Field Encephalography Reconstructed Signal

Building an automatic seizure onset prediction model based on multi-channel electroencephalography (EEG) signals has been a hot topic in computer science and neuroscience field for a long time. In this research, we collect EEG data from different epilepsy patients and EEG devices and reconstruct and combine the EEG signals using an innovative electric field encephalography (EFEG) method, which establishes a virtual electric field vector, enabling extraction of electric field components and increasing detection accuracy compared to the conventional method. We extract a number of important features from the reconstructed signals and pass them through an ensemble model based on support vector machine (SVM), Random Forest (RF), and deep neural network (DNN) classifiers. By applying this EFEG channel combination method, we can achieve the highest detection accuracy at 87% which is 6% to 17% higher than the conventional channel averaging combination method. Meanwhile, to reduce the potential overfitting problem caused by DNN models on a small dataset and limited training patient, we ensemble the DNN model with two “weaker” classifiers to ensure the best performance in model transferring for different patients. Based on these methods, we can achieve the highest detection accuracy at 82% on a new patient using a different EEG device. Thus, we believe our method has good potential to be applied on different commercial and clinical devices.

Application on Anomaly Detection of Geoelectric Field Based on Deep Learning

The deep learning method has made nurnerials achievements regarding anomaly detection in the field of time series.We introduce the speech production model in the field of artificial intelligence,changing the convolution layer of the general convolution neural network to the residual element structure by adding identity mapping,and expanding the receptive domain of the model by using the dilated causal convolution.Based on the dilated causal convolution network and the method of log probability density function,the anomalous events are detected according to the anomaly scores.The validity of the method is verified by the simulation data,which is applied to the actual observed data on the observation staion of Pingliang geoeletric field in Gansu Province.The results show that one month before the Wenchuan M_S8.0,Lushan M_S7.0 and Minxian-Zhangxian M_S6.6 earthquakes,the daily cumulative error of log probability density of the predicted results in Pingliang Station suddenly decreases,which is consistent with the actual earthquake anomalies in a certain time range.After analyzing the combined factors including the spatial electromagnetic environment and the variation of micro fissures before the earthquake,we explain the possible causes of the anomalies in the geoelectric field of before the earthquake.The successful application of deep learning in observed data of the geoelectric field may behefit for improving the ultilization rate both the data and the efficiency of detection.Besides,it may provide technical support for more seismic research.

Medical Knowledge Extraction and Analysis from Electronic Medical Records Using Deep Learning

Objectives Medical knowledge extraction (MKE) plays a key role in natural language processing (NLP) research in electronic medical records (EMR),which are the important digital carriers for recording medical activities of patients.Named entity recognition (NER) and medical relation extraction (MRE) are two basic tasks of MKE.This study aims to improve the recognition accuracy of these two tasks by exploring deep learning methods.Methods This study discussed and built two application scenes of bidirectional long short-term memory combined conditional random field (BiLSTM-CRF) model for NER and MRE tasks.In the data preprocessing of both tasks,a GloVe word embedding model was used to vectorize words.In the NER task,a sequence labeling strategy was used to classify each word tag by the joint probability distribution through the CRF layer.In the MRE task,the medical entity relation category was predicted by transforming the classification problem of a single entity into a sequence classification problem and linking the feature combinations between entities also through the CRF layer.Results Through the validation on the I2B2 2010 public dataset,the BiLSTM-CRF models built in this study got much better results than the baseline methods in the two tasks,where the F1-measure was up to 0.88 in NER task and 0.78 in MRE task.Moreover,the model converged faster and avoided problems such as overfitting.Conclusion This study proved the good performance of deep learning on medical knowledge extraction.It also verified the feasibility of the BiLSTM-CRF model in different application scenarios,laying the foundation for the subsequent work in the EMR field.

English Learning Strategies Employed by EFL Postgraduate Students in Southwest China

English learning strategies play an important role in the language proficiency for EFL learners. Both frequency and patterns of English learning strategies use were found to be significantly related to language proficiency. The present investigation aimed to investigate a) an overall strategy use of postgraduate students in southwest China; and b) to examine the relationship as well as patterns of variations in frequency of students' reported strategy use with reference to their gender and field of study. The subjects of this study were 200 postgraduates in the universities of southwest China. A written questionnaire based on the Strategy Inventory for Language Learning(SILL) was adapted and biodata was added for background information. The findings of the research showed that the postgraduates, on the whole, reported medium frequency of use of English learning strategies. The results of the data analysis also demonstrated that frequency of postgraduates' reported use of English learning strategies varied significantly in terms of gender and field of study. The implications, limitations and recommendations for further research were also discussed.