Evaluation and Prediction of Groundwater Quality in the Municipality of Za-Kpota (South Benin) Using Machine Learning and Remote Sensing

Accessing drinking water is a global issue. This study aims to contribute to the assessment of groundwater quality in the municipality of Za-Kpota (southern Benin) using remote sensing and Machine Learning. The methodological approach used consisted in linking groundwater physico-chemical parameter data collected in the field and in the laboratory using AFNOR 1994 standardized methods to satellite data (Landsat) in order to sketch out a groundwater quality prediction model. The data was processed using QGis (Semi-Automatic Plugin: SCP) and Python (Jupyter Netebook: Prediction) softwares. The results of water analysis from the sampled wells and boreholes indicated that most of the water is acidic (pH varying between 5.59 and 7.83). The water was moderately mineralized, with conductivity values of less than 1500 μs/cm overall (59 µS/cm to 1344 µS/cm), with high concentrations of nitrates and phosphates in places. The dynamics of groundwater quality in the municipality of Za-Kpota between 2008 and 2022 are also marked by a regression in land use units (a regression in vegetation and marshland formation in favor of built-up areas, bare soil, crops and fallow land) revealed by the diachronic analysis of satellite images from 2008, 2013, 2018 and 2022. Surveys of local residents revealed the use of herbicides and pesticides in agricultural fields, which are the main drivers contributing to the groundwater quality deterioration observed in the study area. Field surveys revealed the use of herbicides and pesticides in agricultural fields, which are factors contributing to the deterioration in groundwater quality observed in the study area. The results of the groundwater quality prediction models (ANN, RF and LR) developed led to the conclusion that the model based on Artificial Neural Networks (ANN: R2 = 0.97 and RMSE = 0) is the best for groundwater quality changes modelling in the Za-Kpota municipality.

Accurate and efficient remaining useful life prediction of batteries enabled by physics-informed machine learning

The safe and reliable operation of lithium-ion batteries necessitates the accurate prediction of remaining useful life(RUL).However,this task is challenging due to the diverse ageing mechanisms,various operating conditions,and limited measured signals.Although data-driven methods are perceived as a promising solution,they ignore intrinsic battery physics,leading to compromised accuracy,low efficiency,and low interpretability.In response,this study integrates domain knowledge into deep learning to enhance the RUL prediction performance.We demonstrate accurate RUL prediction using only a single charging curve.First,a generalisable physics-based model is developed to extract ageing-correlated parameters that can describe and explain battery degradation from battery charging data.The parameters inform a deep neural network(DNN)to predict RUL with high accuracy and efficiency.The trained model is validated under 3 types of batteries working under 7 conditions,considering fully charged and partially charged cases.Using data from one cycle only,the proposed method achieves a root mean squared error(RMSE)of 11.42 cycles and a mean absolute relative error(MARE)of 3.19%on average,which are over45%and 44%lower compared to the two state-of-the-art data-driven methods,respectively.Besides its accuracy,the proposed method also outperforms existing methods in terms of efficiency,input burden,and robustness.The inherent relationship between the model parameters and the battery degradation mechanism is further revealed,substantiating the intrinsic superiority of the proposed method.

文本分类中Prompt Learning方法研究综述

文本分类是自然语言处理中的一项基础任务,在情感分析、新闻分类等领域具有重要应用。相较于传统的机器学习和深度学习模型,提示学习可以在数据不足的情况下通过构建提示来进行文本分类。近年来,GPT-3的出现推动了提示学习方法的发展,并且在文本分类领域取得了显著的进展。对以往的文本分类方法进行简要梳理,分析其存在的问题与不足;阐述了提示学习的发展进程,以及构建提示模板的方法,并对用于文本分类的提示学习方法研究及成果进行了介绍和总结。最后,对提示学习在文本分类领域的发展趋势和有待进一步研究的难点进行了总结和展望。

A Hybrid Approach for Predicting the Remaining Useful Life of Bearings Based on the RReliefF Algorithm and Extreme Learning Machine

Accurately predicting the remaining useful life(RUL)of bearings in mining rotating equipment is vital for mining enterprises.This research aims to distinguish the features associated with the RUL of bearings and propose a prediction model based on these selected features.This study proposes a hybrid predictive model to assess the RUL of rolling element bearings.The proposed model begins with the pre-processing of bearing vibration signals to reconstruct sixty time-domain features.The hybrid model selects relevant features from the sixty time-domain features of the vibration signal by adopting the RReliefF feature selection algorithm.Subsequently,the extreme learning machine(ELM)approach is applied to develop a predictive model of RUL based on the optimal features.The model is trained by optimizing its parameters via the grid search approach.The training datasets are adjusted to make them most suitable for the regression model using the cross-validation method.The proposed hybrid model is analyzed and validated using the vibration data taken from the public XJTU-SY rolling element-bearing database.The comparison is constructed with other traditional models.The experimental test results demonstrated that the proposed approach can predict the RUL of bearings with a reliable degree of accuracy.

Machine learning techniques for prediction of capacitance and remaining useful life of supercapacitors: A comprehensive review

Supercapacitors are appealing energy storage devices for their promising features like high power density,outstanding cycling stability,and a quick charge–discharge cycle.The exceptional life cycle and ultimate power capability of supercapacitors are needed in the transportation and renewable energy generation sectors.Hence,predicting the capacitance and lifecycle of supercapacitors is significant for selecting the suitable material and planning replacement intervals for supercapacitors.In addition,system failures can be better addressed by accurately forecasting the lifecycle of SCs.Recently,the use of machine learning for performance prediction of energy storage materials has drawn increasing attention from researchers globally because of its superiority in prediction accuracy,time efficiency,and costeffectiveness.This article presents a detailed review of the progress and advancement of ML techniques for the prediction of capacitance and remaining useful life(RUL)of supercapacitors.The review starts with an introduction to supercapacitor materials and ML applications in energy storage devices,followed by workflow for ML model building for supercapacitor materials.Then,the summary of machine learning applications for the prediction of capacitance and RUL of different supercapacitor materials including EDLCs(carbon based materials),pesudocapacitive(oxides and composites)and hybrid materials is presented.Finally,the general perspective for future directions is also presented.

Deep Learning Applied to Computational Mechanics:A Comprehensive Review,State of the Art,and the Classics

Three recent breakthroughs due to AI in arts and science serve as motivation:An award winning digital image,protein folding,fast matrix multiplication.Many recent developments in artificial neural networks,particularly deep learning(DL),applied and relevant to computational mechanics(solid,fluids,finite-element technology)are reviewed in detail.Both hybrid and pure machine learning(ML)methods are discussed.Hybrid methods combine traditional PDE discretizations with ML methods either(1)to help model complex nonlinear constitutive relations,(2)to nonlinearly reduce the model order for efficient simulation(turbulence),or(3)to accelerate the simulation by predicting certain components in the traditional integration methods.Here,methods(1)and(2)relied on Long-Short-Term Memory(LSTM)architecture,with method(3)relying on convolutional neural networks.Pure ML methods to solve(nonlinear)PDEs are represented by Physics-Informed Neural network(PINN)methods,which could be combined with attention mechanism to address discontinuous solutions.Both LSTM and attention architectures,together with modern and generalized classic optimizers to include stochasticity for DL networks,are extensively reviewed.Kernel machines,including Gaussian processes,are provided to sufficient depth for more advanced works such as shallow networks with infinite width.Not only addressing experts,readers are assumed familiar with computational mechanics,but not with DL,whose concepts and applications are built up from the basics,aiming at bringing first-time learners quickly to the forefront of research.History and limitations of AI are recounted and discussed,with particular attention at pointing out misstatements or misconceptions of the classics,even in well-known references.Positioning and pointing control of a large-deformable beam is given as an example.

Could school programs based on social-emotional learning prevent substance abuse among adolescents?

In this editorial,we comment on the article Adolescent suicide risk factors and the integration of social-emotional skills in school-based prevention programs by Liu et al.While the article focused on the issue of suicide and social-emotional learning programs as a possible intervention,we here discuss evidence of other reported outcomes and if it could be an effective way to prevent substance abuse among adolescents.

The Review of Land Use/Land Cover Mapping AI Methodology and Application in the Era of Remote Sensing Big Data

With the increasing number of remote sensing satellites,the diversification of observation modals,and the continuous advancement of artificial intelligence algorithms,historically opportunities have been brought to the applications of earth observation and information retrieval,including climate change monitoring,natural resource investigation,ecological environment protection,and territorial space planning.Over the past decade,artificial intelligence technology represented by deep learning has made significant contributions to the field of Earth observation.Therefore,this review will focus on the bottlenecks and development process of using deep learning methods for land use/land cover mapping of the Earth’s surface.Firstly,it introduces the basic framework of semantic segmentation network models for land use/land cover mapping.Then,we summarize the development of semantic segmentation models in geographical field,focusing on spatial and semantic feature extraction,context relationship perception,multi-scale effects modelling,and the transferability of models under geographical differences.Then,the application of semantic segmentation models in agricultural management,building boundary extraction,single tree segmentation and inter-species classification are reviewed.Finally,we discuss the future development prospects of deep learning technology in the context of remote sensing big data.

Assessment of glaucoma using extreme learning machine and fractal feature analysis

Dear Sir,Iam Dr.Kavitha S,from the Department of Electronics and Communication Engineering,Nandha Engineering College,Erode,Tamil Nadu,India.I write to present the detection of glaucoma using extreme learning machine(ELM)and fractal feature analysis.Glaucoma is the second most frequent cause of permanent blindness in industrial

Detecting the linkage between arable land use and poverty using machine learning methods at global perspective

Eradicating extreme poverty is one of the UN’s primary sustainable development goals(SDG).Arable land is related to eradicating poverty(SDG1)and hunger(SDG2).However,the linkage between arable land use and poverty reduction is ambiguous and has seldom been investigated globally.Six indicators of agricultural inputs,crop intensification and extensification were used to explore the relationship between arable land use and poverty.Non-parametric machine learning methods were used to analyze the linkage between agriculture and poverty at the global scale,including the classification and regression tree(CART)and random forest models.We found that the yield gap,fertilizer consumption and potential cropland ratio in protected areas correlated with poverty.Developing countries usually had a ratio of actual to potential yield less than 0.33 and fertilizer consumption less than 7.31 kg/ha.Overall,crop extensification,intensification and agricultural inputs were related to poverty at the global level.

GUARDIAN: A Multi-Tiered Defense Architecture for Thwarting Prompt Injection Attacks on LLMs

This paper introduces a novel multi-tiered defense architecture to protect language models from adversarial prompt attacks. We construct adversarial prompts using strategies like role emulation and manipulative assistance to simulate real threats. We introduce a comprehensive, multi-tiered defense framework named GUARDIAN (Guardrails for Upholding Ethics in Language Models) comprising a system prompt filter, pre-processing filter leveraging a toxic classifier and ethical prompt generator, and pre-display filter using the model itself for output screening. Extensive testing on Meta’s Llama-2 model demonstrates the capability to block 100% of attack prompts. The approach also auto-suggests safer prompt alternatives, thereby bolstering language model security. Quantitatively evaluated defense layers and an ethical substitution mechanism represent key innovations to counter sophisticated attacks. The integrated methodology not only fortifies smaller LLMs against emerging cyber threats but also guides the broader application of LLMs in a secure and ethical manner.

Evaluation of a deep-learning model for multispectral remote sensing of land use and crop classification

High-resolution deep-learning-based remote-sensing imagery analysis has been widely used in land-use and crop-classification mapping. However, the influence of composite feature bands, including complex feature indices arising from different sensors on the backbone, patch size, and predictions in transferable deep models require further testing. The experiments were conducted in six sites in Henan province from2019 to 2021. This study sought to enable the transfer of classification models across regions and years for Sentinel-2 A(10-m resolution) and Gaofen PMS(2-m resolution) imagery. With feature selection and up-sampling of small samples, the performance of UNet++ architecture on five backbones and four patch sizes was examined. Joint loss, mean Intersection over Union(m Io U), and epoch time were analyzed, and the optimal backbone and patch size for both sensors were Timm-Reg Net Y-320 and 256 × 256, respectively. The overall accuracy and Fscores of the Sentinel-2 A predictions ranged from 96.86% to 97.72%and 71.29% to 80.75%, respectively, compared to 75.34%–97.72% and 54.89%–73.25% for the Gaofen predictions. The accuracies of each site indicated that patch size exerted a greater influence on model performance than the backbone. The feature-selection-based predictions with UNet++ architecture and upsampling of minor classes demonstrated the capabilities of deep-learning generalization for classifying complex ground objects, offering improved performance compared to the UNet, Deeplab V3+, Random Forest, and Object-Oriented Classification models. In addition to the overall accuracy, confusion matrices,precision, recall, and F1 scores should be evaluated for minor land-cover types. This study contributes to large-scale, dynamic, and near-real-time land-use and crop mapping by integrating deep learning and multi-source remote-sensing imagery.

基于prompt和知识增强的方面级情感分析

方面级情感分析是一种新兴的细粒度情感分析任务,旨在根据给定句子和方面词判断情感极性。目前广泛使用的预训练语言模型由于训练目标和方面级情感分析的目标有差异,分析结果不好。为了缓解预训练语言模型和情感分析目标的差异,prompt被引入到方面级情感分析中,采用伪标签加方面词和意见词的方式创建prompt连续模板,并使用prompt-encoder训练伪标签使其拥有语义信息;然后,使用主题图注意力机制融合关于方面词和意见词的外部知识,根据融合外部知识的隐藏向量预测由情感词典组成的候选标签词;最后,采用求和置信度分数的方式将候选标签词的概率映射到情感极性分布空间上。实验表明,该模型在SemEval 2014任务的笔记本电脑数据集和餐厅数据集上将正确率分别提高了1.53%和3.5%。

The development of machine learning-based remaining useful life prediction for lithium-ion batteries

Lithium-ion batteries are the most widely used energy storage devices,for which the accurate prediction of the remaining useful life(RUL)is crucial to their reliable operation and accident prevention.This work thoroughly investigates the developmental trend of RUL prediction with machine learning(ML)algorithms based on the objective screening and statistics of related papers over the past decade to analyze the research core and find future improvement directions.The possibility of extending lithium-ion battery lifetime using RUL prediction results is also explored in this paper.The ten most used ML algorithms for RUL prediction are first identified in 380 relevant papers.Then the general flow of RUL prediction and an in-depth introduction to the four most used signal pre-processing techniques in RUL prediction are presented.The research core of common ML algorithms is given first time in a uniform format in chronological order.The algorithms are also compared from aspects of accuracy and characteristics comprehensively,and the novel and general improvement directions or opportunities including improvement in early prediction,local regeneration modeling,physical information fusion,generalized transfer learning,and hardware implementation are further outlooked.Finally,the methods of battery lifetime extension are summarized,and the feasibility of using RUL as an indicator for extending battery lifetime is outlooked.Battery lifetime can be extended by optimizing the charging profile serval times according to the accurate RUL prediction results online in the future.This paper aims to give inspiration to the future improvement of ML algorithms in battery RUL prediction and lifetime extension strategy.

Remaining Useful Life Prediction With Partial Sensor Malfunctions Using Deep Adversarial Networks

In recent years,intelligent data-driven prognostic methods have been successfully developed,and good machinery health assessment performance has been achieved through explorations of data from multiple sensors.However,existing datafusion prognostic approaches generally rely on the data availability of all sensors,and are vulnerable to potential sensor malfunctions,which are likely to occur in real industries especially for machines in harsh operating environments.In this paper,a deep learning-based remaining useful life(RUL)prediction method is proposed to address the sensor malfunction problem.A global feature extraction scheme is adopted to fully exploit information of different sensors.Adversarial learning is further introduced to extract generalized sensor-invariant features.Through explorations of both global and shared features,promising and robust RUL prediction performance can be achieved by the proposed method in the testing scenarios with sensor malfunctions.The experimental results suggest the proposed approach is well suited for real industrial applications.

A Hybrid Ensemble Deep Learning Approach for Early Prediction of Battery Remaining Useful Life

Accurate estimation of the remaining useful life(RUL)of lithium-ion batteries is critical for their large-scale deployment as energy storage devices in electric vehicles and stationary storage.A fundamental understanding of the factors affecting RUL is crucial for accelerating battery technology development.However,it is very challenging to predict RUL accurately because of complex degradation mechanisms occurring within the batteries,as well as dynamic operating conditions in practical applications.Moreover,due to insignificant capacity degradation in early stages,early prediction of battery life with early cycle data can be more difficult.In this paper,we propose a hybrid deep learning model for early prediction of battery RUL.The proposed method can effectively combine handcrafted features with domain knowledge and latent features learned by deep networks to boost the performance of RUL early prediction.We also design a non-linear correlation-based method to select effective domain knowledge-based features.Moreover,a novel snapshot ensemble learning strategy is proposed to further enhance model generalization ability without increasing any additional training cost.Our experimental results show that the proposed method not only outperforms other approaches in the primary test set having a similar distribution as the training set,but also generalizes well to the secondary test set having a clearly different distribution with the training set.The PyTorch implementation of our proposed approach is available at https://github.com/batteryrul/battery_rul_early_prediction.

Problem-Based Learning of Drug Use and Abuse during COVID-19 Contingency

Introduction: Nursing students’ experiences during the pandemic provoked social isolation, the way to learn and every context increasing their stress and anxiety leading to drug use and abuse, among others. Problem-based learning (PBL) is a pedagogic strategy to strengthen significant learning;then the objective was to establish PBL influence in nursing students’ experiences on drug use and abuse during COVID-19 contingency. Methods: Qualitative, phenomenological and descriptive paradigm, 12 female and male nursing students aged 20 - 24 years old from the 5th and 6th semesters participated. Information collection was through semi-structured interview and a deep one in four cases. A guide of questions about: How the pandemic impacted your life? How did you face it? And what did you learn during this process? Those questions were used. Qualitative data analysis was based on De Souza Minayo, and signed informed consent was obtained from participants. Results: Students’ experiences allowed four categories to emerge, with six sub-categories. Category I. Students’ experiences on drug use and abuse facing the sanitary contingency;Category II. Students’ skills development to identify a problem and design of appropriate solutions;Category III. Developing skills to favor interpersonal relationships;Category IV. Influence of PBL in nursing students’ experiences on drug use and abuse during the COVID-19 contingency. Conclusion: PBL favored analysis and thoughts in nursing students’ experiences on drug use and abuse during the COVID-19 contingency, they worked collaboratively, developed resilience to daily life situations, and implemented stress coping strategies with their significant learning, which diminished their risk behavior.

融合多Prompt模板的零样本关系抽取模型

Prompt范式被广泛应用于零样本的自然语言处理(NLP)任务中,但是现有基于Prompt范式的零样本关系抽取(RE)模型存在答案空间映射难构造与模板选择依赖人工的问题,无法取得较好的效果。针对这些问题,提出一种融合多Prompt模板的零样本RE模型。首先,将零样本RE任务定义为掩码语言模型(MLM)任务,舍弃答案空间映射的构造,将模板输出的词与关系描述文本在词向量空间中进行比较,以此判断关系类别;其次,引入待抽取关系类别的描述文本的词性作为特征,学习该特征与各个模板之间的权重;最后,利用该权重融合多个模板输出的结果,以此减少人工选取的Prompt模板引起的性能损失。在FewRel(Few-shot Relation extraction dataset)和TACRED(Text Analysis Conference Relation Extraction Dataset)这两个数据集上的实验结果显示,与目前最优的模型RelationPrompt相比,所提模型在不同数据资源设置下,F1值分别提升了1.48~19.84个百分点和15.27~15.75个百分点。可见,所提模型在零样本RE任务上取得了显著的效果提升。

Students＇ Attitude and Their Level of the ICT Use as Learning English Media

This study aims to reveal the students＇ attitude towards the use of Information and Communication Technology （ICT） as learning English Media, to describe the level of their ability to use ICT, and to portray their difficulties to apply Internet at universities. This study used quantitative and qualitative （mixed） method which consisted of two data collection techniques： questionnaire and interview. Respondents total 153 students who responded to the questionnaire and participants interviewed were four students from two private universities in Malang, East Java, Indonesia. The findings show that students have positive attitude towards ICT use as media for learning English. The data analysis strengthens the results which indicate that Intemet as part oflCT functions helps students to elicit useful information. Of the 99.4% participants respond to the questionnaire which stated strongly agree or agree, its mean score is M = 4.69, and its standard deviation is SD = 0.50. The level of students＇ ability to apply ICT is good. Moreover, interview data indicates that students face two difficulties to apply ICT such as Intemet connection problems and ICT illiteracy. The former is regarding the Intemet facility given by the university. The latter deals with the students＇ incompetence in applying ICT because they have no Intemet use experience.

基于Prompt学习的无监督关系抽取模型

无监督关系抽取旨在从无标签的自然语言文本中抽取实体之间的语义关系。目前,基于变分自编码器(VAE)架构的无监督关系抽取模型通过重构损失提供监督信号来训练模型,这为完成无监督关系抽取任务提供了新思路。针对此类模型无法有效地理解上下文信息、依赖数据集归纳偏置的问题,提出基于Prompt学习的无监督关系抽取(PURE)模型,其中包括关系抽取和链接预测两个模块。在关系抽取模块中设计了上下文感知的Prompt模板函数以融入上下文信息,并将无监督关系抽取任务转换为掩码预测任务,从而充分利用预训练阶段获得的知识完成关系抽取。在链接预测模块中则通过预测关系三元组中的缺失实体提供监督信号联合训练两个模块。在两个公开真实关系抽取数据集上进行了大量实验,得到的结果表明PURE模型能有效利用上下文信息并且不依赖数据集归纳偏置,相较于目前最优的基于VAE架构的模型UREVA(Variational Autoencoder-based Unsupervised Relation Extraction model)在NYT数据集上的B-cubed F1指标上提升了3.3个百分点。