Supervised Learning Techniques Research Articles

Machine learning for embodied carbon life cycle assessment of buildings

This paper addresses the significant issue of embodied carbon in buildings and presents a comprehensive approach to its assessment. A machine learning model is proposed, leveraging authentic databases and supervised learning techniques to estimate the environmental impacts of embodied carbon throughout the building life cycle. Validation of the model revealed average percentage errors of approximately 15.71% across different countries. The study also introduces a standardized algorithmic protocol and guidelines for assessing embodied carbon, demonstrated through a case study in Morocco. Results indicate that conventional residential buildings of 120 m2 emit 34.7 tons of embodied carbon, with floors contributing 55%, structure 27%, envelope 14%, and openings 4%. Notably, insulation accounts for 37.0% of the total embodied carbon. Recommendations include incorporating additional databases for learning, considering transportation emissions and primary materials sources, and training the model for different life cycle stages to enhance accuracy. This research provides valuable insights for reducing embodied carbon in buildings and promoting sustainable construction practices.

Economical microscale predictions of wind over complex terrain from mesoscale simulations using machine learning

The ability to assess detailed wind patterns in real time is increasingly important for a variety of applications, including wind energy generation, urban comfort and environmental health, and drone maneuvering in complex environments. Machine Learning techniques are helping to develop accurate and reliable models for predicting local wind patterns. In this paper, we present a method for obtaining wind predictions with a higher resolution, similar to those from computational fluid dynamics (CFD), from coarser, and therefore less expensive, mesoscale predictions of wind in real weather conditions. This is achieved using supervised learning techniques. Four supervised learning approaches are tested: linear regression (SGD), support vector machine (SVM), k-nearest neighbors (KNn) and random forest (RFR). Among the four tested approaches, SVM slightly outperforms the others, with a mean absolute error of 1.81 m/s for wind speed and 40.6∘\\documentclass[12pt]{minimal} \\usepackage{amsmath} \\usepackage{wasysym} \\usepackage{amsfonts} \\usepackage{amssymb} \\usepackage{amsbsy} \\usepackage{mathrsfs} \\usepackage{upgreek} \\setlength{\\oddsidemargin}{-69pt} \\begin{document}$$^{\\circ }$$\\end{document} for wind direction. KNn however achieves the best results in predicting wind direction. Speedup factors of about 290 are achieved by the model with respect to using CFD.

Unraveling the key drivers of community composition in the agri-food trade network

In the complex global food system, the dynamics associated with international food trade have become crucial determinants of food security. In this paper, we employ a community detection approach along with a supervised learning technique to explore the evolution of communities in the agri-food trade network and to identify key factors influencing their composition. By leveraging a large dataset that includes both volume and monetary value of trades, we identify similarities between countries and uncover the primary drivers that shape trade dynamics over time. The analysis also takes into account the impact of evolving climate conditions on food production and trading. The results highlight how the network’s topological structure is continuously evolving, influencing the composition of communities over time. Alongside geographical proximity and geo-political relations, our analysis identifies sustainability, climate and food nutrition aspects as emerging factors that contribute to explaining trade relationships. These findings shed light on the intricate interactions within the global food trade system and provide valuable insights into the factors affecting its stability.

A novel data balancing approach and a deep fractal network with light gradient boosting approach for theft detection in smart grids

Electricity theft is the largest type of non-technical losses faced by power utilities around the globe. It not only raises revenue losses to the utilities but also leads to lethal fires and electric shocks at distribution side. In the past, field operation groups were sent by the utilities to conduct inspections of suspicions electric equipments stated by the public. Advanced metering infrastructure based recent development in the smart grids makes it easy to detect electricity thefts. However, the conventional supervised learning techniques have low theft detection performance mainly due to imbalance datasets available for training. Therefore, in this paper, we develop a novel theft detection model with twofold contribution. A unique hybrid sampling technique named as hybrid oversampling and undersampling using both classes (HOUBC) is proposed to balance the dataset. HOUBC first performs undersampling and then oversampling using both the majority (normal) and minority (theft) classes. A new deep learning method, fractal network is applied with light gradient boosting method to extract and learn important characteristics from electricity consumption profiles for identifying electricity thieves. The proposed model relies on smart meter's data for theft detection and hence, a rapid and widespread adaption of this model is feasible, which shows its main advantage. The performance of the model is evaluated with real-world smart meter's data, i.e., state grid corporation of China. Comprehensive simulation results describe the effectiveness of the proposed model against conventional schemes in terms of electricity theft detection.

A one-class support vector machine for detecting valve stiction

In industrial processes, control valve stiction is known to be one of the primary causes for poor control loop performance. Stiction introduces oscillatory behaviour in the process, leading to increased energy consumption, variations in product quality, shortened equipment lifespan and a reduction in overall plant profitability. Several detection algorithms using routine operating data have been developed over the last few decades. However, with the exception of a handful of recent publications, few attempts to apply classical supervised learning techniques have been published thus far. In this work, principal component analysis, linear discriminant analysis and a one-class support vector machine are trained to detect stiction using time series features as input. These features are extracted from the data using the tsfresh package for Python. The training data consists of simulated stiction examples generated using the XCH stiction model as well as other sources of oscillation. The classifier is subsequently benchmarked against closed-loop stiction data collected in an industrial setting, with performance exceeding that of existing methods.

COVID-19: Symptoms Clustering and Severity Classification Using Machine Learning Approach

COVID-19 is an extremely contagious illness that causes illnesses varying from either the common cold to more chronic illnesses or even death. The constant mutation of a new variant of COVID-19 makes it important to identify the symptom of COVID-19 in order to contain the infection. The use of clustering and classification in machine learning is in mainstream use in different aspects of research, especially in recent years to generate useful knowledge on COVID-19outbreak. Many researchers have shared their COVID-19 data on public database and a lot of studies have been carried out. However, the meritof the dataset is unknown and analysis need to be carried by the researchers to check on its reliability. The dataset that is used in thisworkwas sourced from the Kaggle website. The data wasobtained through a survey collected from participants of various gender and age who had been to at least ten countries.There are four levels of severity based on the COVID-19 symptom, which was developed in accordance to World Health Organization (WHO)and the Indian Ministry of Health and Family Welfare recommendations. This paperpresented an inquiry on the dataset utilising supervised and unsupervised machine learning approaches in order to better comprehend the dataset.In this study, the analysisof the severity group based on theCOVID-19 symptomsusing supervised learning techniques employeda total of seven classifiers, namelythe K-NN, Linear SVM, Naive Bayes, Decision Tree (J48), Ada Boost, Bagging, and Stacking.For the unsupervised learning techniques, the clustering algorithm utilized in this work areSimple K-Means and Expectation-Maximization. From the result obtained from both supervised and unsupervised learning techniques, we observed that the result analysis yielded relatively poor classification and clustering results.The findings for the dataset analysed in this study donot appear to be providing the correctresult for the symptoms categorized against the severity levelwhich raises concerns about the validity and reliability of the dataset.

Deep Learning-Based Attack Detection and Classification in Android Devices

The increasing proliferation of Androidbased devices, which currently dominate the market with a staggering 72% global market share, has made them a prime target for attackers. Consequently, the detection of Android malware has emerged as a critical research area. Both academia and industry have explored various approaches to develop robust and efficient solutions for Android malware detection and classification, yet it remains an ongoing challenge. In this study, we present a supervised learning technique that demonstrates promising results in Android malware detection. The key to our approach lies in the creation of a comprehensive labeled dataset, comprising over 18,000 samples classified into five distinct categories: Adware, Banking, SMS, Riskware, and Benign applications. The effectiveness of our proposed model is validated using well-established datasets such as CICMalDroid2020, CICMalDroid2017, and CICAndMal2017. Comparing our results with state-of-the-art techniques in terms of precision, recall, efficiency, and other relevant factors, our approach outperforms other semi-supervised methods in specific parameters. However, we acknowledge that our model does not exhibit significant deviations when compared to alternative approaches concerning certain aspects. Overall, our research contributes to the ongoing efforts in the development of advanced techniques for Android malware detection and classification. We believe that our findings will inspire further investigations, leading to enhanced security measures and protection for Android devices in the face of evolving threats.

Breast Cancer Detection by Extracting and Selecting Features Using Machine Learning

The cancer of the breast is a significant cause of female death worldwide, but especially in developing countries. For better results and higher survival rates, early diagnosis and screening are crucial. Machine learning (ML) methods can aid in the initialdiscovery and diagnosis of breast cancer by choosing the most informative elements from medical data and eliminating irrelevant ones. The approach of feature extraction involves taking unstructured data and extracting a representative set of characteristics that may be used to classify or forecast data. The aim is to decrease the dimensionality of the feature space while upholding or even refining the accuracy of the ML model. An artificial intelligence model is developed on the given features to categorize mammography images into benign and malignant groups. Different supervised learning techniques, including support vector machines, random forests, and artificial neural networks, are employed and contrasted in order to select the best-performing model. This research offers a comprehensive framework for utilizing machine learning methods to detect breast cancer. The technique demonstrates how it might assist radiologists in the early detection of breast cancer by effectively extracting and selecting critical characteristics that could improve patient outcomes and potentially save lives.

Prediction of Internal Temperature in Greenhouses Using the Supervised Learning Techniques: Linear and Support Vector Regressions

Agricultural greenhouses must accurately predict environmental factors to ensure optimal crop growth and energy management efficiency. However, the existing predictors have limitations when dealing with dynamic, non-linear, and massive temporal data. This study proposes four supervised learning techniques focused on linear regression (LR) and Support Vector Regression (SVR) to predict the internal temperature of a greenhouse. A meteorological station is installed in the greenhouse to collect internal data (temperature, humidity, and dew point) and external data (temperature, humidity, and solar radiation). The data comprises a one year, and is divided into seasons for better analysis and modeling of the internal temperature. The study involves sixteen experiments corresponding to the four models and the four seasons and evaluating the models’ performance using R2, RMSE, MAE, and MAPE metrics, considering an acceptability interval of ±2 °C. The results show that LR models had difficulty maintaining the acceptability interval, while the SVR models adapted to temperature outliers, presenting the highest forecast accuracy among the proposed algorithms.

Low-parameter supervised learning models can discriminate pseudoprogression and true progression in non-perfusion-based MRI.

Discrimination of pseudoprogression and true progression is one challenge to the treatment of malignant gliomas. Although some techniques such as circulating tumor DNA (ctDNA) and perfusion-weighted imaging (PWI) demonstrate promise in distinguishing PsP from TP, we investigate robust and replicable alternatives to distinguish the two entities based on more widely-available media. In this study, we use low-parametric supervised learning techniques based on geographically-weighted regression (GWR) to investigate the utility of both conventional MRI sequences as well as a diffusion-weighted sequence (apparent diffusion coefficient or ADC) in the discrimination of PsP v TP. GWR applied to MRI modality pairs is a unique approach for small sample sizes and is a novel approach in this arena. From our analysis, all modality pairs involving ADC maps, and those involving post-contrast T1 regressed onto T2 showed potential promise. This work on ADC data adds to a growing body of research suggesting the predictive benefits of ADC, and suggests further research on the relationships between post-contrast T1 and T2.Clinical relevance- Few studies have investigated predictive potential of conventional MRI and ADC to detect PsP. Our study adds to the growing research on the topic and presents a new perspective to research by exploiting the utility of ADC in PsP v TP distinction. In addition, our GWR methodology for low-parametric supervised computer vision models demonstrates a unique approach for image processing of small sample sizes.

A2SC: Adversarial Attacks on Subspace Clustering

Many studies demonstrate that supervised learning techniques are vulnerable to adversarial examples. However, adversarial threats in unsupervised learning have not drawn sufficient scholarly attention. In this article, we formally address the unexplored adversarial attacks in the equally important unsupervised clustering field and propose the concept of the adversarial set and adversarial set attack for clustering. To illustrate the basic idea, we design a novel adversarial space-mapping attack algorithm to confuse subspace clustering, one of the mainstream branches of unsupervised clustering. It maps a sample into one wrong class by moving it towards the closest point on the linear subspace of the target class, that is, along the normal of the closest point. This simple single-step algorithm has the power to craft the adversarial set where the image samples can be wrongly clustered, even into the targeted labels. Empirical results on different image datasets verify the effectiveness and superiority of our algorithm. We further show that deep supervised learning algorithms (such as VGG and ResNet) are also vulnerable to our crafted adversarial set, which illustrates the good cross-task transferability of the adversarial set.

Predicting personnel appointment in the Nigerian army using performance records and machine learning techniques

Professionalism and discipline in the Nigerian armed forces have been negatively impacted due to a lack of structured methods of promotion, appointment, and succession in the rank and file of military officers. This lacuna is an attribution of the socio-cultural diversities in Nigeria dispensed through nepotism, favouritism, and ethnicity. Thus, validates the need for pellucid techniques for personnel appointment at the higher echelon based on merit. This paper aims to promote professionalism in the armed forces through a model of seamless human resource processing of enthroning a seamless and transparent culture of succession based on personnel performance records. Supervised learning techniques are adopted for this research given labelled data of 10, 000 records of officers from the rank of major general eligible for appointment as the chief of army staff from the year 1990 to 2002. Relevant features were extracted from the dataset during pre-processing to filter noise, and resampled using sci-kit random over sampler to generate augmented data to balance the target class in order to eliminate algorithmic bias toward the underrepresented class. Three classification algorithms were used comparatively for modelling. The result obtained in terms of accuracy is Logistic regression 84%, decision tree 92%, and random forest 92%. The findings in this research show that our best model random forest will be 92% correct every time prediction is made with a 95 AUC score signifying 95% correctness in distinguishing between the two target classes. This research is the first of its kind and gives room for further improvement with a larger dataset

Applications of artificial intelligence and bioinformatics methodologies in the analysis of ocular biofluid markers: a scoping review.

This scoping review summarizes the applications of artificial intelligence (AI) and bioinformatics methodologies in analysis of ocular biofluid markers. The secondary objective was to explore supervised and unsupervised AI techniques and their predictive accuracies. We also evaluate the integration of bioinformatics with AI tools. This scoping review was conducted across five electronic databases including EMBASE, Medline, Cochrane Central Register of Controlled Trials, Cochrane Database of Systematic Reviews, and Web of Science from inception to July 14, 2021. Studies pertaining to biofluid marker analysis using AI or bioinformatics were included. A total of 10,262 articles were retrieved from all databases and 177 studies met the inclusion criteria. The most commonly studied ocular diseases were diabetic eye diseases, with 50 papers (28%), while glaucoma was explored in 25 studies (14%), age-related macular degeneration in 20 (11%), dry eye disease in 10 (6%), and uveitis in 9 (5%). Supervised learning was used in 91 papers (51%), unsupervised AI in 83 (46%), and bioinformatics in 85 (48%). Ninety-eight papers (55%) used more than one class of AI (e.g. > 1 of supervised, unsupervised, bioinformatics, or statistical techniques), while 79 (45%) used only one. Supervised learning techniques were often used to predict disease status or prognosis, and demonstrated strong accuracy. Unsupervised AI algorithms were used to bolster the accuracy of other algorithms, identify molecularly distinct subgroups, or cluster cases into distinct subgroups that are useful for prediction of the disease course. Finally, bioinformatic tools were used to translate complex biomarker profiles or findings into interpretable data. AI analysis of biofluid markers displayed diagnostic accuracy, provided insight into mechanisms of molecular etiologies, and had the ability to provide individualized targeted therapeutic treatment for patients. Given the progression of AI towards use in both research and the clinic, ophthalmologists should be broadly aware of the commonly used algorithms and their applications. Future research may be aimed at validating algorithms and integrating them in clinical practice.

A REVIEW ON CROP YIELD PREDICTION USING RANDOM FOREST, SVM & KNN

Nowadays we all see machine Learning is famous for data prediction we use Supervised learning techniques & Unsupervised learning techniques. By using this prediction, industries, organizations are improving their performance, productivity & income rate. For supervised learning, we can use algorithms like Random Forests, Support Vector Machines, decision trees & Linear classifiers are used in classification. Linear regression & Logistic regression are used for regression. In unsupervised learning we can use clustering, data mining, etc. in our paper we will collect data on the crop for Maharashtra State for Rabbi & Kharif Season. According to this data set we will apply a supervised technique on it for crop prediction. By using this result farmer can take decision & improve their income rate. Also, we will provide the recent market price of crops to farmers when they use our system for prediction.

IMPROVING CLASSIFICATION PERFORMANCE USING ENSEMBLE LEARNING APPROACH

The data mining techniques are used for evaluation of the data in order to find and represent the data in such manner by which the applications are becomes beneficial. Therefore, different kinds of computational algorithms and modeling’s are incorporated for analyzing the data. These computational algorithms are help to understand the data patterns and their application utility. The data mining algorithms supports supervised as well as unsupervised techniques of data analysis. This work is aimed to investigate about the supervised learning technique specifically performance improvements on classification techniques. The proposed classification model includes the multiple classifiers namely Bayesian classifier, k-nearest neighbor and the c4.5 decision tree algorithm. By nature of the outcomes and the modeling of the data these algorithms are functioning differently from each other. Thus, a weight based classification technique is introduced in this work. The weight is a combination of outcomes provided by the implemented three classifiers in terms of their predicted class labels. Using the weighted outcomes, the final class label for the input data instance is decided. The implementation of the proposed working model is performed with the help of JAVA and WEKA classes. The results obtained by experimentation of the proposed approach with the vehicle data set demonstrate the high accurate classification results. Thus, the proposed model is an effective classification technique as compared to single model implementation for classification task.

Data Mining for Heart Disease Prediction Based on Echocardiogram and Electrocardiogram Data

Traditional methods of detecting cardiac illness are often problematic in the medical field. The doctor must next study and interpret the findings of the patient's medical record received from the electrocardiogram and echocardiogram. These tasks often take a long time and require patience. The use of computational technology in medicine, especially the study of cardiac disease, is not new. Scientists are continuously striving for the most reliable method of diagnosing a patient's cardiac illness, particularly when an integrated system is constructed. The study attempted to propose an alternative for identifying cardiac illness using a supervised learning technique, namely the multi-layer perceptron (MLP). The study started with the collection of patient medical record data, which yielded up to 534 data points, followed by pre-processing and transformation to provide up to 324 data points suitable to be employed by learning algorithms. The last step is to create a heart disease classification model with distinct activation functions using MLP. The degree of classification accuracy, k-fold cross-validation, and bootstrap are all used to test the model. According to the findings of the study, MLP with the Tanh activation function is a more accurate prediction model than logistics and Relu. The classification accuracy level (CA) for MLP with Tanh and k-fold cross-validation is 0.788 in a data-sharing situation, while it is 0.672 with Bootstrap. MLP using the Tanh activation function is the best model based on the CA level and the AUC value, with values of 0.832 (k-fold cross-validation) and 0.857 (bootstrap).

Semi-supervised learning for shale image segmentation with fast normalized cut loss

In analyzing the geological processes, the segmentation of scanning electron microscopy (SEM) images of geological samples is crucial but time-consuming because it needs to distinguish the boundaries for various mineral objects in natural rocks. To automate the segmentation, prior research has adopted supervised learning approaches that train convolutional neural networks (CNNs) using datasets consisting of images and labels. Although supervised learning techniques can produce a high accuracy for a substantial amount of data, the label preparation process is expensive and prone to mistakes because human experts must annotate millions of pixels for each image. To lessen the needs for labeling, in this work we investigated both unsupervised and semi-supervised approaches for fine-grained shale by developing a semi-supervised learning model, SU-Net, based on the U-Net architecture. We also proposed a novel algorithm to speed up the semi-supervised loss function in SU-Net using caching, skip zeros, and batching optimizations. Evaluation results demonstrate that SU-Net can achieve a higher accuracy than U-Net in the case of few labeled data. In addition, SU-Net can be trained as fast as U-Net and is 33 × faster than the state-of-the-art unsupervised model. Overall, SU-Net can benefit the geoscience community by demonstrating its utilities that shale images can be segmented with similar quality even when only limited labels are available.

Micro-hardness evaluation of the bobbin tool-friction stir welded AA6063 using regression-based machine learning

This research work focuses on the prediction of micro-hardness in Bobbin tool friction stir welding (BT-FSW) of AA6063 plates using machine learning (ML) models. The objectives are to determine the effect of process parameters on micro-hardness and to compare the performance of various ML techniques. The dataset obtained from experimental runs is subjected to classical supervised learning techniques such as LR, SVR, DTR, KNN, as well as ensemble techniques including Bagging, Random Forest, Extra Trees, AdaBoost, Gradient Boosting, and Extreme Gradient Boosting. The results show that ML models effectively predict micro-hardness, with ensemble methods outperforming individual techniques. The tool rotational speed is identified as the most influential parameter affecting micro-hardness.

Supervised recommendations of gas metal arc welding parameters

In gas metal arc welding, a weld quality and performance depends on many parameters. Selecting the right ones can be complex, even for an expert. One generally proceeds through trial and error to find a good set of parameters. Therefore, the current experts’ method is not optimized and can require a lot of time and materials. We propose using supervised learning techniques to help experts in their decision-making. To that extent, a two-part recommendation system is proposed. The first step is dedicated to identify, through classification, the number of weld passes. The second one suggests the seven remaining parameter values for each pass: layer, amperage, voltage, wire feed rate, frequency offset, trimming and welding speed. After extracting data from historical Welding Procedure Specification forms, we tested 11 different supervised learning algorithms. The recommendation system is able to provide good results for all the different settings mentioned above even if the data is noisy due to the heuristic nature of the experts’ process. The best classification model is CatBoost with 82.22% average F1 Weighted-Score and the best regression models are Extra Trees or a boosting algorithm with a reduced mean absolute percentage error compared to our baseline.

Multihop Factual Claim Verification Using Natural Language Prompts

Claim verification based on factual evidence can become substantially difficult when the proof comprises several sentences, making it challenging for natural language processing (NLP) models to comprehend long-range dependencies. Inspired by the success of prompt learning in numerous NLP applications, in this paper, we introduce prompt learning for the multi-hop claim verification task. With extensive experiments, we achieve promising results using our proposed prompt-based method that leverages manually constructed prompts. By fine-tuning the language models with prompts, we have achieved an accuracy of 83.9% while obtaining an improved cross-domain generalization performance as well. Moreover, we conducted experiments in few-shot and zero-shot settings and obtained significantly better performance using prompt-based methods compared to traditional supervised learning techniques that are based on the fine-tuning paradigm to further demonstrate the effectiveness of prompt learning in claim verification.