Wrocław University of Technology

About: Wrocław University of Technology is a education organization based out in Wrocław, Poland. It is known for research contribution in the topics: Laser & Fuzzy logic. The organization has 13115 authors who have published 31279 publications receiving 338694 citations.

On detecting and modeling periodic correlation in financial data

Abstract: For many economic problems standard statistical analysis, based on the notion of stationarity, is not adequate. These include modeling seasonal decisions of consumers, forecasting business cycles and—as we show in the present article—modeling wholesale power market prices. We apply standard methods and a novel spectral domain technique to conclude that electricity price returns exhibit periodic correlation with daily and weekly periods. As such they should be modeled with periodically correlated processes. We propose to apply periodic autoregression models which are closely related to the standard instruments in econometric analysis—vector autoregression models.

Azaanalogues of ebselen as antimicrobial and antiviral agents: synthesis and properties.

Abstract: The different analogues of ebselen—unsubstituted benzisoselenazol-3(2H)-one (2a) 2-pyridylbenzisoselenazol-3(2H)-ones (2b–h) and 7-azabenzisoselenazol-3(2H)-ones (3a–j) were designed as new selenium-containing antiviral and antimicrobial agents and synthesized. Some of them were found in the antiviral assay in vitro to be strong inhibitors of cythopatic activity of herpes simplex virus type 1—HSV-1 (compounds 2a,b,f,h, 3a–j) and encephalomyocarditis virus—EMCV (compounds 2a,h, 3a–f,k,l). The compounds 2a,h and 3a–e,j were found to have an appreciable activity against Gram-positive bacteria (Staphylococcus aureus and Bacillus) in vitro, some of them inhibited growth of pathogenic yeasts (Candida albicans) (3a,b) and filamentous fungi (3a–e,f).

Efficient estimation of Markov regime-switching models: An application to electricity spot prices

Abstract: In this paper we discuss the calibration of models built on mean-reverting processes combined with Markov regime-switching (MRS). We propose a method that greatly reduces the computational burden induced by the introduction of independent regimes and perform a simulation study to test its efficiency. Our method allows for a 100 to over 1000 times faster calibration than in case of a competing approach utilizing probabilities of the last 10 observations. It is also more general and admits any value of γ in the base regime dynamics. Since the motivation for this research comes from a recent stream of literature in energy economics, we apply the new method to sample series of electricity spot prices from the German EEX and Australian NSW markets. The proposed MRS models fit these datasets well and replicate the major stylized facts of electricity spot price dynamics.

One-class classifiers with incremental learning and forgetting for data streams with concept drift

Abstract: One of the most important challenges for machine learning community is to develop efficient classifiers which are able to cope with data streams, especially with the presence of the so-called concept drift. This phenomenon is responsible for the change of classification task characteristics, and poses a challenge for the learning model to adapt itself to the current state of the environment. So there is a strong belief that one-class classification is a promising research direction for data stream analysis--it can be used for binary classification without an access to counterexamples, decomposing a multi-class data stream, outlier detection or novel class recognition. This paper reports a novel modification of weighted one-class support vector machine, adapted to the non-stationary streaming data analysis. Our proposition can deal with the gradual concept drift, as the introduced one-class classifier model can adapt its decision boundary to new, incoming data and additionally employs a forgetting mechanism which boosts the ability of the classifier to follow the model changes. In this work, we propose several different strategies for incremental learning and forgetting, and additionally we evaluate them on the basis of several real data streams. Obtained results confirmed the usability of proposed classifier to the problem of data stream classification with the presence of concept drift. Additionally, implemented forgetting mechanism assures the limited memory consumption, because only quite new and valuable examples should be memorized.