New Method Of Machine Learning Research Articles

Time series reconstructing using calibrated reservoir computing

Reservoir computing, a new method of machine learning, has recently been used to predict the state evolution of various chaotic dynamic systems. It has significant advantages in terms of training cost and adjusted parameters; however, the prediction length is limited. For classic reservoir computing, the prediction length can only reach five to six Lyapunov times. Here, we modified the method of reservoir computing by adding feedback, continuous or discrete, to “calibrate” the input of the reservoir and then reconstruct the entire dynamic systems. The reconstruction length appreciably increased and the training length obviously decreased. The reconstructing of dynamical systems is studied in detail under this method. The reconstruction can be significantly improved both in length and accuracy. Additionally, we summarized the effect of different kinds of input feedback. The more it interacts with others in dynamical equations, the better the reconstructions. Nonlinear terms can reveal more information than linear terms once the interaction terms are equal. This method has proven effective via several classical chaotic systems. It can be superior to traditional reservoir computing in reconstruction, provides new hints in computing promotion, and may be used in some real applications.

Machine Learning Methods and Sustainable Development: Metal Oxides and Multilayer Metal Oxides

The development of nanotechnologies and new methods of machine learning are responsible for the significant attention and demand for metal oxides and multilayer metal-oxide nanostructures [...]

Semi-controlled Learning in Information Processing Problems

The article substantiates the need for further research of known methods and the development of new methods of machine learning – semi-supervized learning. It is shown that knowledge of the probability distribution density of the initial data obtained using unlabeled data should carry information useful for deriving the conditional probability distribution density of labels and input data. If this is not the case, semi-supervised learning will not provide any improvement over supervised learning. It may even happen that the use of unlabeled data reduces the accuracy of the prediction. For semi-supervised learning to work, certain assumptions must hold, namely: the semi-supervised smoothness assumption, the clustering assumption (low-density partitioning), and the manifold assumption. A new hybrid semi-supervised learning algorithm using the label propagation method has been developed. An example of using the proposed algorithm is given.

AI based Adaptive Network for Smart Cities

Self-aware, Self-Defending Adaptive Network is a network that defends itself from security breaches in the deployment of smart cities. An adaptive system that knows and recognizes the level of threat faced by an intrusion across a network of smart cities. To detect and separate security threats, the AI program uses a new method of machine learning to track any aspect of a network. Threats include ransomware, high-jacking code, intrusion and illegal entry, theft, and unauthorized use. An autonomous system comprises of a collection of autonomous modules that are introduced and removed dynamically. To order to achieve machine objectives, nodes inside such an ensemble will cooperate. In response to changes in its operating environment, the self-adaptive network modifies its own behavior. We mean anything that the network can observe, such as user interaction, network devices and sensors, or instrumentation, by operating environment.

Modelling of S&P 500 Index Price Based on U.S. Economic Indicators: Machine Learning Approach

In order to forecast stock prices based on economic indicators, many studies have been conducted using well-known statistical methods. Meanwhile, since ~2010 as the power of computers improved, new methods of machine learning began to be used. It would be interesting to know how those algorithms using a variety of mathematical and statistical methods, are able to predict the stock market. The purpose of this article is to model the monthly price of the S&P 500 index based on U.S. economic indicators using statistical, machine learning, deep learning approaches and finally compare metrics of those models. After the selection of indicators according to the data visualization, multicollinearity tests, statistical significance tests, 3 out of 27 indicators remained. The main finding of the research is that the authors improved the baseline statistical linear regression model by 19 percent using a ML Random Forest algorithm. In this way, model achieved accuracy 97.68% of prediction S&P 500 index.

The Evolution of Artificial Intelligence and the Possibility of its Application in Cyber Games

Artificial intelligence, as a separate field of research, is currently experiencing a boom - new methods of machine learning and hardware are emerging and improving, and the results achieved change the life of society. Machine translation, handwriting recognition, speech recognition are changing our reality. The work of creating unmanned vehicles, voice assistants and other devices using these technologies is in an active process. The article examines the historical context of the artificial intelligence development, it evaluates the possibilities of its introduction into cyber games, as a safe and effective platform for testing new methods of machine learning. The promotion of such projects can increase the reputation of development companies, ensure increased user confidence in other products and, with a competent marketing strategy, cause a significant public resonance among video game fans, providing the developer with economic profit.

A Comparative Study of SVM and RF Methods for Classification of Alteration Zones Using Remotely Sensed Data

Identification and mapping of the significant alterations are the main objectives of the exploration geochemical surveys. The field study is time-consuming and costly to produce the classified maps. Therefore, the processing of remotely sensed data, which provide timely and multi-band (multi-layer) data, can be substituted for the field study. In this study, the ASTER imagery is used for alteration classification by applying two new methods of machine learning, including Random Forest and Support Vector Machine. The 14 band ASTER and 19 derivative data layers extracted from ASTER including band ratio and PC imagery, are used as training datasets for improving the results. Comparison of analytical results achieved from the two mentioned methods confirmed that the SVM model has sufficient accuracy and more powerful performance than RF model for alteration classification in the study area.

From monitoring of seismic fields to the automatic forecasting of earthquakes

PurposeThe purpose of this paper is to offer two Web-based platforms for systematic analysis of seismic processes. Both platforms are designed to analyze and forecast the state of the environment and, in particular, the level of seismic hazard. The first platform analyzes the fields representing the properties of the seismic process; the second platform forecasts strong earthquakes. Earthquake forecasting is based on a new one-class classification method.Design/methodology/approachThe paper suggests an approach to systematic forecasting of earthquakes and examines the results of tests. This approach is based on a new method of machine learning, called the method of the minimum area of alarm. The method allows to construct a forecast rule that optimizes the probability of detecting target earthquakes in a learning sample set, provided that the area of the alarm zone does not exceed a predetermined one.FindingsThe paper presents two platforms alongside the method of analysis. It was shown that these platforms can be used for systematic analysis of seismic process. By testing of the earthquake forecasting method in several regions, it was shown that the method of the minimum area of alarm has satisfactory forecast quality.Originality/valueThe described technology has two advantages: simplicity of configuration for a new problem area and a combination of interactive easy analysis supported by intuitive operations and a simplified user interface with a detailed, comprehensive analysis of spatio-temporal processes intended for specialists. The method of the minimum area of alarm solves the problem of one-class classification. The method is original. It uses in training the precedents of anomalous objects and statistically takes into account normal objects.

A study of decision tree application in the problem of accounting for non-insured periods of a pensioner

The article is devoted to the study of machine learning algorithm “decision Tree” for solving the problem “Accounting for non-insured periods”. In pension legislation, there are problems that required a large amount of combinatorial calculations. These include the problem of “Accounting for non-insured periods”. To solve similar problems, machine learning methods should be suitable. To implement the algorithm, training and test samples are created by means of a random number sensor. However, studies have shown that the use of this method gives satisfactory results, but not acceptable for this task. The authors conclude that it is necessary to investigate other, more complex machine learning algorithms, or to find a new dependence of the output variable on the input variables. This dependence is probably to be nonlinear that will determine a choice of a new method of machine learning.

Multivariable least squares support vector machine with time integral operator for the prediction of bearing performance degradation

The prediction of performance degradation is significant for the health monitoring of rolling bearing, which helps to greatly reduce the loss caused by potential faults in the entire life cycle of rotating machinery. As a new method of machine learning based on statistical learning theory, least squares support vector machine is developed and has achieved good results. However, it lacks the description of the time-sum effect and delay characteristics, which cannot fully describe the performance degradation process. To overcome the problem, a new time shift least squares support vector machine with integral operator is proposed. What is more, multivariable prediction model is introduced to describe the process from multiple perspectives. In this model, different features are extracted to construct sample pairs through a moving window. Then these features are decomposed in time domain using a set of orthogonal basis functions to simplify computation. Furthermore, the model adaptability is also improved through an iterative updating strategy. Bearing fault experiments show that the proposed model outperforms the general method.

A Deep Learning Approach to the Classification of 3D Models under BIM Environment

In recent years, the extensive use of BIM has brought a revolutionary change for the construction industry. As an important technology to support BIM, three-dimensional digital technology has become a hot academic research. The 3D geometric model is the main data expression of modes in BIM environment. But because of the complexity of 3D modeling, the maintenance of the 3D model library in BIM environment will spend a lot of time and cost. The traditional CAD 3D modeling has accumulated a large number of 3D models for the BIM project to reuse. Using the 3D model classification technology can quickly classify the existing 3D model, and save a lot of cost. Deep learning is in recent years a new method of machine learning. In this paper, we use Stacked Auto-Encoders (SAE) to classify 3D models under the environment of BIM. Experiments show that, the method proposed in the paper has achieved good results in the 3D model classification, which provides a new idea for the development of BIM.

A New Method of Machine Learning Based on Examples

A new method of machine learning based on examples is given in this paper. This method improves the classical method ID3 which learns from static examples. Its limits lie on no comprehension and no memory, and no dynamic correlation. In the new method, it can learn from dynamic examples, the change of data can be learned because the training data is the initial and end process in the interval. All varieties and correlation can be understood and remembered. By experiments, the method can be used as classifier and it has special use in the field of information mining.

A multiscale feature fusion approach for classification of very high resolution satellite imagery based on wavelet transform

A novel methodology based on multiscale spectral and spatial information fusion using wavelet transform is proposed in order to classify very high resolution (VHR) satellite imagery. Conventional wavelet‐based feature extraction methods employ single windows of a fixed size, which are not satisfactory as the VHR imagery contains complex and multiscale objects. In this paper, spectral and spatial features are extracted based on a set of concentric windows around a central pixel in order to integrate the information across different windows/scales. The proposed method is made up of three blocks: (1) the conventional wavelet‐based feature extraction methods are extended from single band processing to multispectral bands, and from single window to multi‐windows, (2) two multiscale fusion algorithms are proposed to exploit the multiscale spectral and spatial information and (3) a support vector machine (SVM), a relatively new method of machine learning, is used to classify the multiscale spectral–spatial feature sets. The proposed classification method is evaluated on two VHR datasets and the results show that the multiscale approach can improve the classification accuracy in homogeneous areas while simultaneously preserving accuracy in edge regions.

Application of Machine Learning to Short-Term Equity Return Prediction

Cooper showed how a filter method could be used to predict equity returns for the next week by using information about returns and volume for the two previous weeks. Cooper's method may be regarded as a crude method of Machine Learning. Over the last 20 years Machine Learning has been successfully applied to the modeling of large data sets, often containing a lot of noise, in many different fields. When applying the technique it is important to fit it to the specific problem under consideration. We have designed and applied to Cooper's problem a practical new method of Machine Learning, appropriate to the problem, that is based on a modification of the well-known kernel regression method. We call it the Prototype Kernel Regression method (PKR). In both the period 1978-1993 studied by Cooper, and the period 1994-2004, the PKR method leads to a clear profit improvement compared to Cooper's approach. In all of 48 different cases studied, the period pre-cost average return is larger for the PKR method than Cooper's method, on average 37% higher, and that margin would increase as costs were taken into account. Our method aims to minimize the danger of data snooping, and it could plausibly have been applied in 1994 or earlier. There may be a lesson here for proponents of the Efficient Market Hypothesis in the form that states that profitable prediction of equity returns is impossible except by chance. It is not enough for them to show that the profits from an anomaly-based trading scheme disappear after costs. The proponents should also consider what would have been plausible applications of more sophisticated Machine Learning techniques before dismissing evidence against the EMH.