Unsupervised and Supervised Learning Combined Power Load Curve Classification Based on Sequential Trajectory Feature Extraction Algorithm

Abstract. Hyperspectral images (HSIs) contain hundreds of spectral bands, providing high-resolution spectral information pertaining to the Earth’s surface. Additionally, abundant spatial contextual information can also be obtained simultaneously from a HSI. To characterize the properties of ground objects, classification is the most widely-used technology in the field of remote sensing, where each pixel in a HSI is assigned to a pre-defined class. Over the past decade, deep learning has attracted increasing attention in the machine-learning and computer-vision domains, due to its favourable performances for various types of tasks, and it has been successfully introduced to the remote-sensing community. Instead of utilizing the shallow features within in a given image, which is the approach that is generally adopted in other conventional classification methods, deep-learning algorithms can extract hierarchical features from raw HSI data. Within the deep-learning framework, recurrent neural networks (RNNs), which are able to encode sequential features, have exhibited promising capabilities and have achieved encouraging performances, especially for the natural-language processing and speech-recognition communities. As multi-temporal remote-sensing images can be readily obtained from increasing numbers of satellite and unmanned aircraft systems, and since analysis of such multi-temporal data comprises a critical issue within numerous research subfields, including land-cover and land-change analyses, and land-resource management, RNNs have been applied in recent studies in order to extract temporal sequential features from multi-temporal remote-sensing images for the purpose of image classification. Apart from using multi-temporal image datasets, RNNs can also be utilized on a single image, where the spectral feature/band of each individual pixel can be taken as a sequential feature for the input layer of RNNs. However, the application of such sequential feature extraction that relies on a single image still needs to be further investigated since applying RNNs to spectral bands will directly introduce more parameters that need to be optimized, consequently increasing the total training time.In this study, we propose a novel RNN-based HSI classification framework. In this framework, unlabelled pixels obtained from a single image are considered when constructing sequential features. Two spatial similarity measurements, referred to as pixel-matching and block-matching, respectively, are employed to extract pixels that are “similar” to the target pixel. Then, the sequential feature of the target pixel is constructed by exploiting several of the most “similar” pixels and ordering them based on their similarities to the target pixel. The aforementioned two schemes are advantageous, as unlabelled pixels within the given HSI are taken into consideration for similarity measurement and sequential feature construction for the RNN model. Moreover, the block-matching scheme also takes advantage of spatial contextual information, which has been widely utilized in spatial-spectral-based HSI classification methods. To evaluate the proposed methods, two benchmark HSIs are used, including a HSI collected over Pavia University, Italy by the airborne Reflective Optics System Imaging Spectrometer (ROSIS) sensor, and an image acquired over the Salinas Valley, California, USA via the Airborne Visible/Infrared Imaging Spectrometer (AVIRIS) sensor. Spatio-temporally coincident ground-reference data accompanies each of these respective HSIs. In addition, the proposed methods are compared with three state-of-the-art algorithms, including support vector machine (SVM), the 1-dimensional convolutional neural network (1DCNN), and the 1-dimensional RNN (1DRNN).Experimental results indicate that our proposed methods achieve markedly better classification performance compared with the baseline algorithms on both datasets. For example, for the Pavia University image, the block-matching based RNN achieves the highest overall classification accuracy, with 94.32% accuracy, which is 9.87% higher than the next most accurate algorithm of the aforementioned three baseline methods, which in this case is the 1DCNN, with 84.45% overall accuracy. More specifically, the block-matching method performs better than the pixel-matching method in terms of both quantitative and qualitative assessments. Based on visual assessment/interpretation of the classification maps, it is apparent that “salt-and-pepper” noise is markedly alleviated; with block-matching, smoother classified images are generated compared with pixel-matching-based methods and the three baseline algorithms. Such results demonstrate the effectiveness of utilizing spatial contextual information in the similarity measurement.

Skin disease is a very vulnerable and severe issue in today's world. Skin disorder classification is crucial for diagnosis. Several new data mining algorithms have been developed to classify and interpret medical images. The functionality of the K-Nearest Neighbors (KNN) Random Forest (RF) Algorithm is described in this article, along with an analysis of its results. Furthermore, this study demonstrates a high-performing approach that saves both effort and money. The proposed model is designed based on KNN and the Random Forest algorithm. The patient can use this model to classify his skin disease as a primary detection, and the doctor also can ensure his judgment by using this proposed model. Traditional skin disease diagnosis is an expensive and time-consuming procedure. This paper's proposed classification model will identify ten different skin diseases. The Random Forest algorithm has a testing accuracy of 94.22 percent, and K-Nearest Neighbors (KNN) has a testing accuracy of 95.23 percent. The KNN algorithm has an F1 Score of 95.98 percent, whereas the Random Forest (RF) algorithm has an F1 Score of 95.94 percent. It can be increased by expanding the dataset and more feature extraction. This approach may benefit individuals with skin illness who are looking to save money and time as well as avoid skin cancer by identifying cancer at an early stage.

The random forest(RF) algorithm is a very efficient and excellent ensemble classification algorithm. In this paper, we improve the random forest algorithm and propose an algorithm called ‘Bayesian Weighted Random Forest’(B-RF), focus on the problem that inaccurate decision tree classification caused by the same voting weights in the traditional random forest model. The main idea underlying the proposed model is to replace the supermajority voting of random forests into weighted voting, fully consider the difference of classification ability of each decision tree, using the Bayesian formula to dynamically update the weight value for each tree, so that the strong classifier has higher voting power and effectively improves the overall performance of classification. Through the verification of UCI database, the results show that the classification accuracy of the proposed weighted random forest model is higher. This illustrate the outperformance of the proposed model in comparison with the general random forest algorithm.

Context.As an increasing number of spectroscopic surveys become available, an automated approach to spectral classification becomes necessary. Due to the significance of the massive stars, it is of great importance to identify the phenomenological parameters of these stars (e.g., the spectral type), which can be used as proxies to their physical parameters (e.g., mass and temperature).Aims.In this work, we aim to use the random forest (RF) algorithm to develop a tool for the automated spectral classification of OB-type stars according to their sub-types.Methods.We used the regular RF algorithm, the probabilistic RF, which is an extension of RF that incorporates uncertainties, and we introduced the KDE – RF method which is a combination of the kernel-density estimation and the RF algorithm. We trained the algorithms on the equivalent width (EW) of characteristic absorption lines measured in high-quality spectra (signal-to-noise (S/N)≳50) from large Galactic (LAMOST, GOSSS) and extragalactic surveys (2dF, VFTS) with available spectral types and luminosity classes. By following an adaptive binning approach, we grouped the labels of these data in 11 spectral classes within the O2-B9 range. We examined which of the characteristic spectral lines (features) are more important for the classification based on a number of feature selection methods, and we searched for the optimal hyperparameters of the classifiers to achieve the best performance.Results.From the feature-screening process, we find that the full set of 17 spectral lines is needed to reach the maximum performance per spectral class. We find that the overall accuracy score is ∼70%, with similar results across all approaches. We apply our model in other observational data sets providing examples of the potential application of our classifier to real science cases. We find that it performs well for both single massive stars and for the companion massive stars in Be X-ray binaries, especially for data of similar quality to the training sample. In addition, we propose a reduced ten-features scheme that can be applied to large data sets with lowerS/N ∼ 20 − 50.Conclusions.The similarity in the performances of our models indicates the robustness and the reliability of the RF algorithm when it is used for the spectral classification of early-type stars. The score of ∼70% is high if we consider (a) the complexity of such multiclass classification problems (i.e., 11 classes), (b) the intrinsic scatter of the EW distributions within the examined spectral classes, and (c) the diversity of the training set since we use data obtained from different surveys with different observing strategies. In addition, the approach presented in this work is applicable to products from different surveys in terms of quality (e.g., different resolution) and different formats (e.g., absolute or normalized flux), while our classifier is agnostic to the luminosity class of a star, and, as much as possible, it is metallicity independent.

Application of random forest algorithm in hail forecasting over Shandong Peninsula

Fusion of diverse data sources for flood extent mapping and risk assessment in Sindh: A comparative study of inundation mapping approaches

Urban areas represent the primary source region of greenhouse gas emissions. Mapping urban areas is essential for understanding land cover change, carbon cycles, and climate change (urban areas also refer to impervious surfaces, i.e., artificial cover and structures). Remote sensing has greatly advanced urban areas mapping over the last several decades. At present, we have entered the era of big data. Long time series of satellite data such as Landsat and high-performance computing platforms such as Google Earth Engine (GEE) offer new opportunities to map urban areas. The objective of this research was to determine how annual time series images from Landsat 8 Operational Land Imager (OLI) can effectively be composed to map urban areas in three cities in China in support of GEE. Three reducer functions, ee.Reducer.min(), ee.Reducer.median(), and ee.Reducer.max() provided by GEE, were selected to construct four schemes to synthesize the annual intensive time series Landsat 8 OLI data for three cities in China. Then, urban areas were mapped based on the random forest algorithm and the accuracy was evaluated in detail. The results show that (1) the quality of annual composite images was improved significantly, particularly in reducing the impact of cloud and cloud shadows, and (2) the annual composite images obtained by the combination of multiple reducer functions had better performance than that obtained by a single reducer function. Further, the overall accuracy of urban areas mapping with the combination of multiple reducer functions exceeded 90% in all three cities in China. In summary, a suitable combination of reducer functions for synthesizing annual time series images can enhance data quality and ensure differences between characteristics and higher precision for urban areas mapping.

In this paper, a semi-supervised random forest (RF) algorithm is presented for fault diagnosis of rotating machinery. Firstly, many unlabeled samples are divided into two parts, denoted respectively as unlabeled sample I and unlabeled sample II. Then a graph-all the labeled samples are used to train the multiple decision trees. If the classification result is consistent with the one of label prediction, then the unlabeled sample I is added to the labeled samples and used for building RF model. While, the data of unlabeled sample II are utilized for testing of the obtained RF model. Finally, the developed RF algorithm is applied to an experimental platform of rotating machinery. It is shown from the simulation results that, for the cases of noisy samples with unsatisfying labels, the new developed semi-supervised RF algorithm can improve the fault classification accuracy than the conventional RF.

The advancing information technology is playing more and more important role in data mining of relational database.1 The transfer and sharing of databases cause the copyright-related security threats. Database watermarking technology can effectively solve the problem with copyright protection and traceability, which has been attracting researchers' attention. In this paper, we proposed a novel, robust and reversible database watermarking technique, named histogram shifting watermarking based on random forest and genetic algorithm (RF-GAHCSW). It greatly improves the watermark capacity by means of histogram width reduction and eliminates the impact of the prediction error attack. Meanwhile, random forest algorithm is used to select important attributes for watermark embedding, and genetic algorithm is employed to find the optimal secret key for the database grouping and determine the position of watermark embedding to improve the watermark capacity and reduce data distortion. The experimental results show that the robustness of RF-GAHCSW is greatly improved, compared with the original HSW, and the distortion has little effect on the usability of database.

A spatially based machine learning algorithm for potential mapping of the hearing senses in an urban environment

Along with the steady growth of wired and wireless networks, the various new attacks targeting networks are also constantly emerging and transforming. As a efficient way to cope with various attacks, the Random Forest(RF) algorithm has frequently been used as the core engine of intrusion detection because of the faster learning speed and the higher attack detection accuracy. However, the RF algorithm has to input the number of the tree composing the forest as a parameter. In this paper, we proposed a new algorithm that limit the number of trees composing the forest using the McNemar test. To evaluate the performance of the proposed RF algorithm, we compared learning time, accuracy and memory usage of the proposed algorithm with the original RF algorithm and other algorithm by using the KDDcup99 dataset. Under the same detection accuracy, the proposed RF algorithm improves the performance of the original RF algorithm by as much as 97.76% at learning time, 91.86% at test time, and 99.02% in memory usage on average.

Accurate estimation of dew point temperature (T_dew) plays a very important role in the fields of water resource management, agricultural engineering, climatology and energy utilization. However, there are few studies on the applicability of local T_dew algorithms at regional scales. This study evaluated the performance of a new machine learning algorithm, i.e., gradient boosting on decision trees with categorical features support (CatBoost) to estimate daily T_dew using limited local and cross-station meteorological data. The random forests (RF) algorithm was also assessed for comparison. Daily meteorological data from 2016 to 2019, including maximum, minimum and average temperature (T_max, T_min and T_mean), maximum, minimum and average relative humidity (RH_max, RH_min and RH_mean), maximum, minimum and average global solar radiation (Rs_max, Rs_min and Rs_mean) from three weather stations in Hunan of China were used to evaluate the CatBoost and RF algorithms. The results showed that both algorithms achieved satisfactory estimation accuracy at the target stations (on average RMSE = 1.020°C, R²=0.969, MAE = 0.718°C and NRMSE = 0.087) in the absence of complete meteorological parameters (with only temperature data as input). The CatBoost algorithm (on average RMSE = 1.900°C and R²=0.835) was better than the RF algorithm (on average RMSE = 2.214°C and R²=0.828). The accuracy and stability of the CatBoost and RF algorithms were positively correlated with the number of input parameters, and the three-parameter algorithms achieved higher estimation accuracy than the two-parameter algorithms. The developed methodology is helpful to predict T_dew at regional scale.

Lameness is a serious disease that affects the health and welfare of dairy cattle whilst also causing yield and economic losses. The primary goal of this study is to determine if lameness can be detected early on in herd management using the Random Forest (RF) algorithm and the surface temperatures of the cows' hoof soles, as well as the digital colour parameters generated by processing these thermal camera images. Ages, hoof sole temperatures, and digital colour characteristics of 40 Simmental cattle were used as independent variables in this study, while lameness was evaluated by scoring and employed as a dependent variable after being updated as a binary variable. The parameters ntree= 100 and mtry= 3 were used to develop the RF algorithm for predicting lameness in animals. As a result, the RF algorithm correctly classified 19 of 22 healthy animals and incorrectly classified 3, while it correctly classified 15 of 18 unhealthy animals and incorrectly classified 3. The classification success of the RF algorithm was 85%, sensitivity, specificity and area under the ROC curve (AUC) were 0.864, 0.833, and 0.848±0.059, respectively, and it was successful in detecting lameness. Also, AUC, which is one of the RF algorithm's classification performances, was found to be statistically significant (P<0.05). As a direct consequence it can be stated that the RF algorithm is a suitable classifier in terms of the use of animal hoof sole temperatures and digital colour parameters obtained through image processing in the detection of lameness in herd management.

Combinations of oxaliplatin combined with various schedules of antimetabolite 5- fluorouracil and leucovorin (FOLFOX) have been the first-line treatment for unresectable colorectal cancer (CRC) and survival benefits for the patients undergoing FOLFOX therapy were demonstrated. That is, the median duration of survival for the group treated with FOLFOX has been reported to be about 20 months, compared with 8 months for the best supportive care group. However, the rate of effect is not so high and has been reported to be approximately 50%. Molecular characterization using gene expression profiling will undoubtedly improve the prediction of treatment responses and ultimately the clinical outcome of cancer patients, which has never been accomplished by conventional clinicopathological classification. The aim of this study was to identify responders for FOLFOX therapy by applying Random Forests (RF) algorithm to gene expression data. Sixty-six unresectable colorectal cancer (CRC) patients were included, which were divided into training and test sets. Evaluated by computed tomography after 4 courses of FOLFOX therapy, 25 patients were diagnosed as responders and 25 patients as non-responders, respectively, in 50 training samples. Using RF, 22 of 25 responders (88.0% sensitivity) and 24 of 25 non-responders (96.0% specificity) were correctly classified, with an accuracy of 92.0% in out-of-bag cross-validation, and predictor genes were identified for FOLFOX therapy. Based on the predictor genes, proximities could clearly classify tumor samples into three subtypes, responders, non-responders, and outliers. After getting 5 outliers out of 50 training samples, RF was performed again with identification of predictor genes, such as RDX, SMURF2, and PEG10. In order to validate these predictor genes, we applied these decision tree analyses to independent 16 test samples, and achieved 75.9% sensitivity, 88.9% specificity, and 84.6% of classification accuracy. Based on these predictor genes, we performed Gene Set Enrichment Analysis and found activated pathways such as ‘hematopoietic stem cell genes’ pathway (including CIRBP and NPEPPS). Furthermore, we investigated survival benefit of the patients who were diagnosed as responders for FOLFOX therapy by RF algorithm and found the survival distribution for unresectable CRC outcome. Taken together, RF on gene expression data for unresectable CRC patients could effectively stratify responders for FOLFOX therapy with high accuracy rate, and pharmacogenomics in anticancer therapy is definitely the first step for personalized therapy. Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 102nd Annual Meeting of the American Association for Cancer Research; 2011 Apr 2-6; Orlando, FL. Philadelphia (PA): AACR; Cancer Res 2011;71(8 Suppl):Abstract nr 4135. doi:10.1158/1538-7445.AM2011-4135

The development of digital technology has brought significant changes across various sectors of life, including transportation. One of the most popular modes of transportation among the public today is online motorcycle taxis, such as Gojek. Gojek continues to innovate to meet customer needs more effectively and to expand its range of services. This study aims to identify the number of positive, neutral, and negative sentiments in a user review dataset, as well as to evaluate the performance of the algorithms used—namely, SVM and Random Forest. The analysis was conducted on 10,000 customer reviews from the Play Store application, resulting in 2,057 positive sentiments, 1,135 neutral sentiments, and 6,295 negative sentiments. The classification model compared the SVM algorithm with the Random Forest algorithm, and the results show that Random Forest achieved better performance, with 91% accuracy compared to SVM’s 89%. These findings demonstrate that Random Forest performs better in handling word distribution within review texts than the SVM method.