Regio (Estonia)

In this paper, we present classical machine learning algorithms enhanced by quantum technology to classify a data set. The data set contains binary input variables and binary output variables. The goal is to extend classical approaches such as neural networks by using quantum machine learning principles. Classical algorithms struggle as the dimensionality of the feature space increases. We examine the usage of quantum technologies to speed up these classical algorithms and to introduce the new quantum paradigm into machine diagnostic domain. Most of the prognosis models based on binary or multi-valued classification have become increasingly complex throughout the recent past. Starting with a short introduction into quantum computing, we will present an approach of combining quantum computing and classical machine learning to the reader. Further, we show different implementations of quantum classification algorithms. Finally, the algorithms presented are applied to a data set. The results of the implementations are shown and discussed in the last section.

This paper proposes a technique for improving the accuracy of mobile device movement trajectory reconstruction using passive mobile positioning data. The major sources of uncertainty in trajectory reconstruction are imprecise cell shape data and ‘ping-pong’ effects caused by cell handovers. We used a novel technique for improved ‘ping-pong’ effect suppression by compensating for some cell shape distortions based on temporal cell-to-cell transit statistics. The results were evaluated by estimating traffic flow using trajectory reconstruction. The proposed technique improved the accuracy of results compared to ‘ping-pong’ suppression algorithms found in the literature.