Jerzy Świątek

System analysis techniques in ehealth systems: a case study

In the paper problem of planning training protocol with taking into account limitations on the training intensity due to the health problems of the exerciser is considered. In the first part of the work short introduction to existing solutions in the area of eHealth applications is given. Next, architecture of the eHealth system to support exerciser training is discussed. The main functionalities of proposed system are pointed out and challenges are highlighted. The concept of context-awareness and personalization is stressed. At the end the problem of model based optimisation of the training protocol is formulated.

Ensemble Classifier for Solving Credit Scoring Problems

The goal of this paper is to propose an ensemble classification method for the credit assignment problem. The idea of the proposed method is based on switching class labels techniques. An application of such techniques allows solving two typical data mining problems: a predicament of imbalanced dataset, and an issue of asymmetric cost matrix. The performance of the proposed solution is evaluated on German Credits dataset.

Static and dynamic complex models: comparison and application to chemical systems

Purpose – The purpose of this paper is to present research in the area of the modeling of complex systems using feed‐forward neural network.Design/methodology/approach – Applications of multilayer neural networks with supervisor learning on the own simulator program wrote in Borland® Pascal Language. Series‐parallel identification method is applied. Tapped delay lines (TDL) in static neural networks for modeling of dynamic plants are used. Gradient and heuristic learning algorithms are applied. Three kinds of calibration of learning and testing data are used.Findings – This paper illustrates that feed‐forward multilayer neural networks can model complex systems. Feed‐forward multilayer neural networks with TDL can be used to build global dynamic models of complex systems. It is possible to compare the quality both models.Research limitations/implications – The learning and testing data from real systems to tune neuronal models require use of calibrating these data to range 0‐1.Practical implications – The...

Personalisation in Service-Oriented Systems Using Markov Chain Model and Bayesian Inference

In the paper a personalization method using Markov model and Bayesian inference is presented. The idea is based on the hypothesis that user’s choice of a new decision is influenced by the last made decision. Thus, the user’s behaviour could be described by the Markov chain model. The extracted knowledge about users’ behaviour is maintained in the transition matrice as probability distribution functions. An estimation of probabilities is made by applying incremental learning algorithm which allows to cope with evolving environments (e.g. preferences). At the end an empirical study is given. The proposed approach is presented on an example of students enrolling to courses. The dataset is partially based on real-life data taken from Wroclaw University of Technology and includes evolving users’ behaviour.

Accelerated learning for Restricted Boltzmann Machine with momentum term

Restricted Boltzmann Machines are generative models which can be used as standalone feature extractors, or as a parameter initialization for deeper models. Typically, these models are trained using Contrastive Divergence algorithm, an approximation of the stochastic gradient descent method. In this paper, we aim at speeding up the convergence of the learning procedure by applying the momentum method and the Nesterov’s accelerated gradient technique. We evaluate these two techniques empirically using the image dataset MNIST.

Application of Nonlinear State Estimation Methods for Sport Training Support

Typical understanding of healthcare concerns treatment, diagnosis and monitoring of diseases. But healthcare also includes well-being, healthy lifestyle, and maintaining good body condition. One of the most important factor in this respect is physical activity. Modern techniques of data acquisition and data processing enable development of advanced systems for physical activity support with use of measurement data. The need for reliable estimation routines stems from the fact, that many widely available for bulk customers measurements devices are not reliable and measured signals are contaminated by the noise. One of the most important variables for physical activity monitoring is the velocity of a moving object e.g. velocity of selected parts of a body such as elbows. Apart from intensive use of system identification, optimization and control techniques for physical training support, we applied Kalman filtering technique in order to estimate speed of moving part of a body.

Cost Sensitive SVM with Non-informative Examples Elimination for Imbalanced Postoperative Risk Management Problem

In this paper we propose a novel combined approach to solve the imbalanced data issue in the application to the problem of the post-operative life expectancy prediction for the lung cancer patients. This solution makes use of undersampling techniques together with cost-sensitive SVM (Support Vector Machines). First, we eliminate non-informative examples by applying Tomek links together with one-sided selection. Second, we take advantage of using cost-sensitive SVM with penalty costs calculated respecting cardinalities of minority and majority examples. We evaluate the presented solution by comparing the performance of our method with SVM-based approaches that deal with uneven data. The experimental evaluation was performed on real-life data from the postoperative risk management domain.

Computational Investigation of Probabilistic Learning Task with Use of Machine Learning

Probabilistic Learning Task is a game that serve psychiatrists and psychologists to measure some cognitive abilities of people having various cognitive disorders. Mathematical models together with machine learning techniques are routinely used to summarize large amount of data produced by players during the game. Parameters of mathematical models are taken to represent behavioral data gathered during the game. However, there is no study of reliability of those parameters available in literature. We investigate how much one can trust the values of models parameters. We proposed a specific method to assess reliability of models parameters, that makes use of the game sessions of human players and their virtual counterparts.

Self-paced Learning for Imbalanced Data

In this paper, we propose a novel training paradigm that combines two learning strategies: cost-sensitive and self-paced learning. This learning approach can be applied to the decision problems where highly imbalanced data is used during training process. The main idea behind the proposed method is to start the learning process by taking large number of minority examples and only the easiest majority objects and then gradually turning to more difficult cases. We examine the quality of this training paradigm comparing to other learning schemas for neural network model using a set of highly imbalanced benchmark datasets.

Test Scenarios and Validation Results of POP Methodology

The paper presents the results of quantitative and qualitative assessment of effectiveness of POP Methodology (a selection of agile methods of analysis, planning and optimization of business process management information systems, developed by a team of researchers at PWr). In particular, the POP Methodology evaluation metrics are proposed, focusing mainly on the process of software product development and the product quality assessment in various stages of the process. Factors such as duration, experience and size of a development team are considered. The proposed metrics are discussed and illustrated based on a few selected examples.