Search results

It is well known that the selection of initial cluster centers can significantly affect K-means clustering results. The purpose of this paper is to propose an improved, efficient methodology to handle such a challenge.

According to the fact that the inner-class distance among samples within the same cluster is supposed to be smaller than the inter-class distance among clusters, the algorithm will dynamically adjust initial cluster centers that are randomly selected. Consequently, such adjusted initial cluster centers will be highly representative in the sense that they are distributed among as many samples as possible. As a result, local optima that are common in K-means clustering can then be effectively reduced. In addition, the algorithm is able to obtain all initial cluster centers simultaneously (instead of one center at a time) during the dynamic adjustment.

Experimental results demonstrate that the proposed algorithm greatly improves the accuracy of traditional K-means clustering results and, in a more efficient manner.

The authors presented in this paper an efficient algorithm, which is able to dynamically adjust initial cluster centers that are randomly selected. The adjusted centers are highly representative, i.e. they are distributed among as many samples as possible. As a result, local optima that are common in K-means clustering can be effectively reduced so that the authors can achieve an improved clustering accuracy. In addition, the algorithm is a cost-efficient one and the enhanced clustering accuracy can be obtained in a more efficient manner compared with traditional K-means algorithm.

This paper aims to introduce the take-off and landing performance analysis modules of the software library named Java toolchain of Programs for Aircraft Design (JPAD), dedicated to the aircraft preliminary design. An overview of JPAD is also presented.

The calculation of the take-off and landing distances has been implemented using a simulation-based approach. This expects to solve an appropriate set of ordinary differential equations, which describes the aircraft equations of motion during all the take-off and landing phases. Tests upon two aircraft models (ATR72 and B747-100B) have been performed to compare the obtained output with the performance data retrieved from the related flight manuals.

The tool developed has proven to be very reliable and versatile, as it performs the calculation of the required performance with almost no computational effort and with a good accuracy, providing a less than the 5 per cent difference with respect to the statistical trend and a difference from the flight manual or public brochure data around 10 per cent.

The use of a simulation-based approach to have a more accurate estimation of the ground performance with respect to classic semi-empirical equations. Although performing the simulation of the aircraft motion, the approach shown is very time-saving and can be easily implemented in an optimization cycle.

The purpose of this paper is geared towards the capture and analysis of network traffic using an array ofmachine learning (ML) and deep learning (DL) techniques to classify network traffic into different classes and predict network traffic parameters.

The classifier models include k-nearest neighbour (KNN), multilayer perceptron (MLP) and support vector machine (SVM), while the regression models studied are multiple linear regression (MLR) as well as MLP. The analytics were performed on both a local server and a servlet hosted on the international business machines cloud. Moreover, the local server could aggregate data from multiple devices on the network and perform collaborative ML to predict network parameters. With optimised hyperparameters, analytical models were incorporated in the cloud hosted Java servlets that operate on a client–server basis where the back-end communicates with Cloudant databases.

Regarding classification, it was found that KNN performs significantly better than MLP and SVM with a comparative precision gain of approximately 7%, when classifying both Wi-Fi and long term evolution (LTE) traffic.

Collaborative regression models using traffic collected from two devices were experimented and resulted in an increased average accuracy of 0.50% for all variables, with a multivariate MLP model.