Search

Machine learning-based classification algorithms allow communication and computing (2C) task offloading from the end devices to the edge computing network servers. In this paper, we consider task classification based on the hybrid k-means and k ′ -nearest neighbors algorithms. Moreover, we examine the poisoning attacks on such ML algorithms, namely noise-like jamming and targeted data feature falsification, and their impact on the effectiveness of 2C task allocation. Then, we also present two anomaly detection methods using noise training and the silhouette score test to detect the poisoned samples and mitigate their impact. Our simulation results show that these attacks have a fatal effect on classification in feature areas where the decision boundary is unclear. They also demonstrate the effectiveness of our countermeasures against the considered attacks. [ABSTRACT FROM AUTHOR]

Optical "fingerprints" are widely used for chemometrics-assisted recognition of samples of different types. An emerging trend in this area is the transition from obtaining "static" spectral data to reactions analyzed over time. Indicator reactions are usually carried out in aqueous solutions; in this study, we developed reactions that proceed in an organic solvent, thereby making it possible to recognize fat-soluble samples. In this capacity, we used 5W40, 10W40, and 5W30 motor oils from four manufacturers, with six samples in total. The procedure involved mixing a dye, sample, and reagents (HNO3, HCl, or tert-butyl hydroperoxide) in an ethanolic solution in a 96-well plate and measuring absorbance or near-infrared fluorescence intensity every several minutes for 20–55 min. The obtained photographic images were processed by linear discriminant analysis (LDA) and the k-nearest neighbors algorithm (kNN). Discrimination accuracy was evaluated by a validation procedure. A reaction of oxidation of a dye by nitric acid allowed us to recognize all six samples with 100% accuracy for LDA. Merging of data from the four reactions that did not provide complete discrimination ensured an accuracy of 93% for kNN. The newly developed indicator systems have good prospects for the discrimination of other fat-soluble samples. Overall, the results confirm the viability of the kinetics-based discrimination strategy. [ABSTRACT FROM AUTHOR]

Due to the limited number of air quality monitoring stations, the data collected are limited. Using supervised learning for air quality fine-grained analysis, that is used to predict the air quality index (AQI) of the locations without air quality monitoring stations, may lead to overfitting in that the models have superior performance on the training set but perform poorly on the validation and testing set. In order to avoid this problem in supervised learning, the most effective solution is to increase the amount of data, but in this study, this is not realistic. Fortunately, semi-supervised learning can obtain knowledge from unlabeled samples, thus solving the problem caused by insufficient training samples. Therefore, a co-training semi-supervised learning method combining the K-nearest neighbors (KNN) algorithm and deep neural network (DNN) is proposed, named KNN-DNN, which makes full use of unlabeled samples to improve the model performance for fine-grained air quality analysis. Temperature, humidity, the concentrations of pollutants and source type are used as input variables, and the KNN algorithm and DNN model are used as learners. For each learner, the labeled data are used as the initial training set to model the relationship between the input variables and the AQI. In the iterative process, by labeling the unlabeled samples, a pseudo-sample with the highest confidence is selected to expand the training set. The proposed model is evaluated on a real dataset collected by monitoring stations from 1 February to 30 April 2018 over a region between 118° E–118°53′ E and 39°45′ N–39°89′ N. Practical application shows that the proposed model has a significant effect on the fine-grained analysis of air quality. The coefficient of determination between the predicted value and the true value is 0.97, which is better than other models. [ABSTRACT FROM AUTHOR]

Three different types of entropy weight methods (EWMs), i.e., EWM-A, EWM-B, and EWM-C, have been used by previous studies for integrating prediction models. These three methods use very different ideas on determining the weights of individual models for integration. To evaluate the performances of these three EWMs, this study applied them to developing integrated short-term traffic flow prediction models for signalized intersections. At first, two individual models, i.e., a k-nearest neighbors (KNN)-algorithm-based model and a neural-network-based model (Elman), were developed as individual models to be integrated using EWMs. These two models were selected because they have been widely used for traffic flow prediction and have been approved to be able to achieve good performance. After that, three integrated models were developed by using the three different types of EWMs. The performances of the three integrated models, as well as the individual KNN and Elman models, were compared. We found that the traffic flow predicted with the EWM-C model is the most accurate prediction for most of the days. Based on the model evaluation results, the advantages of using the EWM-C method were deliberated and the problems with the EWM-A and EWM-B methods were also discussed. [ABSTRACT FROM AUTHOR]

Micro-plasma transfer arc metal additive manufacturing (μ-PTAMAM) process is a unique direct energy deposition (DED) type additive manufacturing (AM) process capable of manufacturing three-dimensional metallic components through simultaneous use of wire and powder forms of AM material. DED process yields an uneven surface on the manufactured components which necessitates their further post-processing to attain the required surface finish, dimensional accuracy, and geometrical accuracy. This research focuses on using machine learning algorithm, namely, K-nearest neighbors (KNN), to predict the surface roughness. The data of surface roughness for training KNN algorithm was generated by depositing multi-layer single-track depositions yielding wall-like structures using Stellite-6 as AM material in powder and wire forms. It was found that surface roughness increases with an increase in power supply to micro-plasma and AM material feed rate whereas it decreases with an increase in traverse speed of the deposition head for both powder and wire of the AM material. Surface roughness of the walls for powder form of the AM material is smaller (i.e., 118 to 149 μm) than that obtained by the wire form (i.e., 195 to 227 μm). Surface roughness prediction error for KNN algorithm is found to be from − 6.2 to 2.8% for powder form and − 5.8 to 2.3% for wire form of the AM material proving capability of KNN algorithm for accurate prediction of surface roughness produced by μ-PTAMAM process. Since the prediction error depends on number of data set used to train KNN algorithm, therefore, it can be further reduced by increasing number of training data set. [ABSTRACT FROM AUTHOR]

Food irradiation is becoming increasingly popular in many countries for preserving and extending the shelf life of foods, which creates a demand for express methods for the detection of absorbed doses. This paper proposes an innovative method for the estimation of the dose absorbed by X-rayed beef samples using a reaction-based optical sensing technique that was proposed earlier by our team for the estimation of the dose absorbed by raw potatoes. Potato and beef samples were exposed to X-ray irradiation at 100 and 1000 Gy and then extracted by water at 23 °C for 24 h or 60/70 °C for 1 h. The resulting solutions were introduced to the reaction mixtures of dyes (carbocyanines, Rhodamines, or Crystal Violet) with oxidants (hypochlorite, bromate, or hydrogen peroxide). The fluorescence intensity and absorbance of the mixtures were periodically measured photographically using visualizers. The data were processed using linear discriminant analysis (LDA) and k-nearest neighbors algorithm (kNN). Using the most efficient individual reactions, the doses can be recognized with a 90–100% accuracy, and the combinations of up to 5 reactions can improve the accuracy to 100%. The protocol is simple and rapid: sample extraction time and indicator reactions take not more than 1 h each. The proposed method potentially has a wide area of applications ranging from plants to animal products. • A simple chemical method for estimating doses is developed. • No prior knowledge about irradiation markers is required. • 1-Hour extraction of irradiated samples with water is sufficient. • 100% accuracy of dose estimation (0, 100, 1000 Gy) is achieved. [ABSTRACT FROM AUTHOR]

In financial data mining, stock index movement direction prediction is a challenging classification problem, since stock index is affected by many economic and political factors. The accurate prediction of this problem is of interest to many researchers as it can serve as an early recommender system for short-term financiers. This study aims to predict daily upward or downward movement direction of Borsa Istanbul 100 (XU100) index with the aid of supervised machine learning algorithms based on classification. Problem we deal with includes whether on a specific day the XU100 index fall into up bucket or fall into down bucket. For this purpose, the multiple logistic regression and Knearest neighbors algorithm models are fitted using independent variables whose effect on XU100 index movement direction was statistically significant. Lastly, the out-of sample predictions are compared with the actual movements in the stock market. Performances are measured not only with accuracy but also other statistical metrics. According to the results obtained, logistic regression analysis achieves better predict performance with 81% accuracy opposed to K-nearest neighbors algorithm on XU100 data over the given time period. [ABSTRACT FROM AUTHOR]

An automatic defect classification (ADC) system identifies and classifies wafer surface defects using scanning electron microscope images. By classifying defects, manufacturers can determine whether the wafer can be repaired and proceed to the next fabrication step. Current ADC systems have high defect detection performance. However, the classification power is poor. In most work sites, defect classification is performed manually using the naked eye, which is unreliable. This paper proposes an ADC method based on deep learning that automatically classifies various types of wafer surface damage. In contrast to conventional ADC methods, which apply a series of image recognition and machine learning techniques to find features for defect classification, the proposed model adopts a single convolutional neural network (CNN) model that can extract effective features for defect classification without using additional feature extraction algorithms. Moreover, the proposed method can identify defect classes not seen during training by comparing the CNN features of the unseen classes with those of the trained classes. Experiments with real datasets verified that the proposed ADC method achieves high defect classification performance. [ABSTRACT FROM AUTHOR]

This paper aims to do a comparative study of local feature descriptor techniques and convolutional neural networks (CNN) for retrieving Thai silk pattern images. Two feature descriptor techniques, the histogram of oriented gradients and the scale-invariant feature transform, are compared to extract feature vectors from the silk pattern images. We combined the feature vectors extracted from feature descriptor techniques with k-nearest neighbors (KNN) and support vector machine. Then we modified CNN architectures: LeNet-5 and AlexNet. The LeNet-5 was modified by increasing the number of neurons in each layer of the fully connected layers. The AlexNet architecture was modified by reducing the neurons in each layer of the fully connected layers. Finally, we evaluated the local descriptor techniques, the existing CNN architectures and our modified CNN architectures on Thai silk pattern dataset. The results of the study showed that the local descriptor techniques combined with KNN algorithm significantly outperform the CNN methods. [ABSTRACT FROM AUTHOR]

In this paper, we propose a local descriptors approach to classify pornographic images. Two local descriptors including the scale-invariant feature transform (SIFT) and the histogram of oriented gradients (HOG) are computed feature vectors from pornographic images. The extracted features are supplied to the K-Nearest Neighbor Algorithm and Support Vector Machine for pornographic image classification. We have evaluated all different methods on the TI-UNRAM dataset. The results show that the HOG and the SIFT combined with the SVM significantly outperform the other methods, including KNN algorithm and image processing technique. Thus, the local descriptors approach can be combined with the SVM for pornographic image classification in order to obtain the effective classification performance. [ABSTRACT FROM AUTHOR]

K-nearest neighbors (KNN) algorithm is a common algorithm used for classification, and also a sub-routine in various complicated machine learning tasks. In this paper, we presented a quantum algorithm (QKNN) for implementing this algorithm based on the metric of Hamming distance. We put forward a quantum circuit for computing Hamming distance between testing sample and each feature vector in the training set. Taking advantage of this method, we realized a good analog for classical KNN algorithm by setting a distance threshold value t to select k − n e a r e s t neighbors. As a result, QKNN achieves O( n ) performance which is only relevant to the dimension of feature vectors and high classification accuracy, outperforms Llyod's algorithm (Lloyd et al. 2013) and Wiebe's algorithm (Wiebe et al. 2014). [ABSTRACT FROM AUTHOR]

Breast cancer, has been a significant cancer type for women on the society. Early diagnosis and timely medical treatment are important key factors spreading to the other tissues and permitting long-time survival of patients. Since the existing methods have several serious shortfalls, microwave imaging method for the diagnosis of early stage tumors has been interested by different scientific research groups in terms of moderating endogenous the electrical property difference between healthy tissue and malignancies. In this article, both an ultra-wideband bow-tie antenna with enhanced bandwidth and a 3D breast model which has different electrical properties which are permittivity and conductivity is created in simulation tool to solve electromagnetic field values. Return loss, VSWR, and radiation pattern characteristics, which are significant antenna parameters, are simulated and obtained whether the antenna possess an efficient characteristic or not. Electric field values over the breast tissue in which there is a tumor or not tumor are evaluated. In this article, above-mentioned values of frequency bandwidth, dielectric constant of antenna's substrate, electric field, and tumor information were consisted in their dataset. This dataset obtained from the Bow-Tie Antenna was used to detect the breast cancer with one of the data mining method, which is K-Nearest Neighbor Algorithm. [ABSTRACT FROM AUTHOR]

A possibility of identification of the oxidation state of iron by wavelength dispersive X-ray fluorescence spectroscopy using both the position and intensity of L and L spectral lines of iron and by principal component analysis score data obtained by the decomposition of the spectral region corresponding to spectral L-series lines of iron is demonstrated. The application of scores ensures a more reliable identification in comparison with line parameters (position and intensity). Two approaches based on projection on latent structures ( PLS) regression for the determination of the concentration of iron in different oxidation states are proposed. The first approach consists in using reference models with compositions similar to those of analyzed samples. In the second approach, PLS regression was build using model spectra obtained from spectra of readily available iron compounds. [ABSTRACT FROM AUTHOR]

The use of/(-Nearest Neighbors (/cNN) algorithm for recognizing handwritten character scripts is presented in this survey paper. The kNN algorithm, the simple and well-known algorithm for machine learning, is suitable for classification scheme. According to the/cNN algorithm, unknown data is firstly compared with the training samples to compute the similarity function. The most similar k neighbors, which have the smallest distance values, are subsequently selected by using the majority vote. The unknown data is then assigned by determining the number of its occurrences in each class. The most frequent class in k neighbors is finally selected as the output of the classifier. เท this survey, it is found that the handwritten character datasets are collected from the different types of handwriting, including different character sets with the similar structure, a variety of the handwriting styles, incomplete handwritten characters and insufficient number of handwritten character samples. These factors have a directl impact on the performance of the handwritten character recognition. Thus, the robust feature extraction techniques for handwritten recognition can be combined with the/cNN algorithm in order to obtain the effective recognition performance. [ABSTRACT FROM AUTHOR]

The topic logistic has become more and more important in German agriculture during the last years. This is caused by a growth of enterprises and machines but also by the enormous extension of the cultivation of renewable resources for the production of energy. To manage these logistical tasks in agriculture in Germany at the moment, different transport systems are preferred. The classical system with tractor and agricultural trailer, transport via truck like it, is typical for the commercial transport of goods and the transport with specialized vehicles which can be classified between both systems. To evaluate these transport processes it is decisive for the farmers to know the key parameters of the single systems like the average fuel consumption (energy) and the average transport speed (time) for their special logistic issue to optimize the input of resources. The aim of the examination is therefore to develop a planning tool for the farmers to evaluate the logistic systems. With in this project examinations have shown that environmental influences like driver, loading, road type, gradient, winding ness, traffic situation, daytime and so on, have an enormous influence on the key parameters energy and time. Current simulation systems for the estimation of the key parameters energy and time have not provided satisfying results for the agricultural sector. Usually they are basing on linear and/or nonlinear equations with a very complex emphasis of the influencing factors and after technical changes on the logistic systems, they cannot be used any longer. Furthermore always only a very small number of influencing factors can be integrated. Therefore in this examination the k-Nearest Neighbors algorithm (kNN) is used. This results in a much more flexible use of different influencing factors with a difference in weight. With the help of test data, the system is learning and creating the kNN algorithm. This can be used for simulation. The advantage of this system of "artificial intelligence" is that the model building can be done in time in the area of the current working point. This makes it possible to integrate even unknown or in their effect not determinable environmental factors. By the training structure and the integration of new test data, the algorithm is much more easily adaptable on new trends. There is also the possibility to train the system optimal on the own conditions with the help of own test data. The k-Nearest Neighbors algorithm which has been determined during the examination, makes it possible, to estimate the key parameters energy and time for the logistic tasks in agriculture with a probability of more than 97%. [ABSTRACT FROM AUTHOR]

The article concerns the problem of classification based on independent data sets—local decision tables. The aim of the paper is to propose a classification model for dispersed data using a modified k-nearest neighbors algorithm and a neural network. A neural network, more specifically a multilayer perceptron, is used to combine the prediction results obtained based on local tables. Prediction results are stored in the measurement level and generated using a modified k-nearest neighbors algorithm. The task of neural networks is to combine these results and provide a common prediction. In the article various structures of neural networks (different number of neurons in the hidden layer) are studied and the results are compared with the results generated by other fusion methods, such as the majority voting, the Borda count method, the sum rule, the method that is based on decision templates and the method that is based on theory of evidence. Based on the obtained results, it was found that the neural network always generates unambiguous decisions, which is a great advantage as most of the other fusion methods generate ties. Moreover, if only unambiguous results were considered, the use of a neural network gives much better results than other fusion methods. If we allow ambiguity, some fusion methods are slightly better, but it is the result of this fact that it is possible to generate few decisions for the test object. [ABSTRACT FROM AUTHOR]

As resources allocated specifically for conservation are limited, there is a need to ensure conservation policy initiatives lead to effective conservation outcomes. In this study, we investigated the potential conservation benefits from alternative spatial allocations of old deciduous stands to a landscape dominated by coniferous production forests owned primarily by non-industrial private forest owners. As a target species, we used the long-tailed tit ( Aegithalos caudatus), a species associated with deciduous forests and known to be sensitive to isolation. We used a previously published model based on empirical data on the occurrence of this species, to assess the probability of occurrence of the bird in a 4,000 km area in southern Sweden for which we possess detailed spatial GIS data (kNN data) of tree species composition and age. We assessed alternative scenarios where old deciduous forest was allocated with or without respect to distance from existing old deciduous forests. Due to the long-tailed tit's habitat requirement increasing the amount of old deciduous forests close to existing habitats was the most effective strategy. However, the potential advantages of this strategy may in fact be overturned in favor of the other scenarios if ownership structures and probable uptake rates of policy initiatives are also considered. If a policy initiative is targeted toward owners with properties in close proximity to existing suitable habitat, when compared to if all forest owners are targeted, a higher proportion of owners is needed to participate in order to achieve the same degree of habitat creation for the species. Here, we discuss the potential benefits for effective conservation policy formulation from integrating spatially explicit datasets and detailed ecological knowledge with land-ownership structures and policy uptake scenarios. [ABSTRACT FROM AUTHOR]

This paper presents a novel approach for an Intrusion Detection System (IDS) based on one kind of asymmetric optimization which use any three already well-known IDS algorithms and Triple Modular Redundancy (TMR) algorithm together. Namely, a variable threshold which indicates an attack on an observed and protected network is determined by using all three values obtained with three known IDS algorithms i.e., on previously recorded data by making a decision by majority. For these algorithms authors used algorithm of k-nearest neighbors, cumulative sum algorithm, and algorithm of exponentially weighted moving average. Using a proposed method we can get a threshold that is more precisely determined than in the case of any method individual. Practically, using TMR we obtain a dynamically threshold adjustment of IDS software, which reduces the existence of false alarms and undetected attacks, so the efficiency of such IDS software is notably higher and can get better results. Today, Denial of Service attacks (DoS) are one of the most present type of attacks and the reason for the special attention paid to them in this paper. In addition, the authors of the proposed method for IDS software used a known CIC-DDoS2019 dataset, which contains various data recordings of such attacks. Obtained results with the proposed solution showed better characteristics than each individual used algorithm in this solution. IDS software with the proposed method worked precisely and timely, which means alarms were triggered properly and efficiently. [ABSTRACT FROM AUTHOR]

(1) Background: In this work, we focus on the activity of large-conductance voltage- and Ca 2 + -activated potassium channels (BK) from the inner mitochondrial membrane (mitoBK). The characteristic electrophysiological features of the mitoBK channels are relatively high single-channel conductance (ca. 300 pS) and types of activating and deactivating stimuli. Nevertheless, depending on the isoformal composition of mitoBK channels in a given membrane patch and the type of auxiliary regulatory subunits (which can be co-assembled to the mitoBK channel protein) the characteristics of conformational dynamics of the channel protein can be altered. Consequently, the individual features of experimental series describing single-channel activity obtained by patch-clamp method can also vary. (2) Methods: Artificial intelligence approaches (deep learning) were used to classify the patch-clamp outputs of mitoBK activity from different cell types. (3) Results: Application of the K-nearest neighbors algorithm (KNN) and the autoencoder neural network allowed to perform the classification of the electrophysiological signals with a very good accuracy, which indicates that the conformational dynamics of the analyzed mitoBK channels from different cell types significantly differs. (4) Conclusion: We displayed the utility of machine-learning methodology in the research of ion channel gating, even in cases when the behavior of very similar microbiosystems is analyzed. A short excerpt from the patch-clamp recording can serve as a "fingerprint" used to recognize the mitoBK gating dynamics in the patches of membrane from different cell types. [ABSTRACT FROM AUTHOR]