Tomasz Orczyk

The k-NN classifier and self-adaptive Hotelling data reduction technique in handwritten signatures recognition

The paper proposes a novel signature verification concept. This new approach uses appropriate similarity coefficients to evaluate the associations between the signature features. This association, called the new composed feature, enables the calculation of a new form of similarity between objects. The most important advantage of the proposed solution is case-by-case matching of similarity coefficients to a signature features, which can be utilized to assess whether a given signature is genuine or forged. The procedure, as described, has been repeated for each person presented in a signatures database. In the verification stage, a two-class classifier recognizes genuine and forged signatures. In this paper, a broad range of classifiers are evaluated. These classifiers all operate on features observed and computed during the data preparation stage. The set of signature composed features of a given person can be reduced what decrease verification error. Such a phenomenon does not occur for the raw features. The approach proposed was tested in a practical environment, with handwritten signatures used as the objects to be compared. The high level of signature recognition obtained confirms that the proposed methodology is efficient and that it can be adapted to accommodate as yet unknown features. The approach proposed can be incorporated into biometric systems.

Fingerprint ridges frequency

This work is a study about the fingerprint ridge distance and presents an efficient algorithm to estimate the values of local ridge frequency. Most of fingerprint image enhancement algorithms use an anisotropic filters with frequency response. Therefore, ridges frequency, as a global feature of fingerprint, is very important to image preprocessing methods used in automatic fingerprint identification system (AFIS). In proposed algorithm, through the use of fingerprint direction image, the frequency values are calculated directly, without the use of computationally expensive spectral analysis.

DTW and voting-based lip print recognition system

This paper presents a method of lip print comparison and recognition. In the first stage the appropriate lip print features are extracted. Lip prints can be captured by police departments. Traces from lips may also found at a crime scene. The approach uses the well known DTW algorithm and Copeland vote counting method. Tests were conducted on 120 lip print images. The results obtained are very promising and suggest that the proposed recognition method can be introduced into professional forensic identification systems.

Fusion of Granular Computing and k–NN Classifiers for Medical Data Support System

The medical data and its classification should be particularly treated. The data can not be modified or altered, because this could lead to overestimation or false decisions. Some classifiers, using random factors, can generate false, higher overall accuracy of diagnosis. Medical support systems should be trustworthy and reliable even at the cost of system complexity. In this paper fusion of two classifiers has been proposed, where k–NN classifier and classifier based on a justified granulation paradigm were employed. Additionally, proposed solution allows to visualize obtained classification results. Accuracy of the proposed solution has been compared with various classifiers. All methods presented in this work were tested on real medical data coming from three medical datasets. Finally, some remarks for further research have been proposed.

Liver Fibrosis Diagnosis Support System Using Machine Learning Methods

Liver fibrosis is a common disease of the European population (but not only them). It may have many backgrounds and may develop with a different rapidity—it may stay hidden for many years or rapidly develop into terminal stage called cirrhosis, where liver can no longer fulfill its function. Unfortunately, current methods of diagnosis are either connected with a potential risk for a patient and require a hospitalization or are expensive and not very accurate. This paper presents a comparative study of various feature selection algorithms combined with selected machine learning algorithms which may be used to build an advanced liver fibrosis diagnosis support system based on a nonexpensive and safe routine blood tests. Experiments carried out on a dataset collected by authors, proved usability and satisfactory accuracy of the presented algorithms.

Investigation of the Impact of Missing Value Imputation Methods on the k-NN Classification Accuracy

Paper desribes results of an experiment where various scenarios of missing values occurrence in the data repository has been tested. Experiment was coducted on a publicly available database, containing complete, multidimensional continuous dataspace and multiple classes. Missing values were introduced using “completely at random” scheme. Tested scenarios were: training and testing using incomplete dataset, training on complete data set and testing on incomplete and vice versa. For comparison to data imputation methods also the ensemble of single-feature kNN classifiers, working withoud data imputation, has been tested.

Cost Sensitive Hierarchical Classifiers for Non-invasive Recognition of Liver Fibrosis Stage

Liver Fibrosis caused by the Hepatitis Virus type C (HCV) may be a serious life-threatening condition if is not diagnosed and treated on time. Our previous research proved that it is possible to estimate liver fibrosis stage in patients with diagnosed HCV only using blood tests. The aim of our research is to find a safe and non-invasive but also inexpensive diagnostic method. As not all blood tests are equally expensive (not only in meaning of money, but also time of analysis), this article introduces a Cost Factor to the hierarchical classifiers. Our classifier has been based on a C4.5 decision tree building algorithm enhanced with a modified EG2 algorithm for maintaining a cost limit.

Adaptive splitting and selection method for noninvasive recognition of liver fibrosis stage

Therapy of patients suffer form liver diseases strongly depends on the liver fibrosis progression. Unfortunately, to asses it the liver biopsy has been usually used which is an invasive and raging medical procedure which could lead to serious health complications. Additionally even when experienced medical experts perform liver biopsy and read the findings, up to a 20% error rate in liver fibrosis staging has been reported. Nowadays a few noninvasive commercial tests based on the blood examinations are available for the mentioned above problem. Unfortunately they are quite expensive and usually they are not refundable by the health insurance in Poland. Thus, the cross-disciplinary team, which includes researches form the Polish medical and technical universities has started work on new noninvasive method of liver fibrosis stage classification. This paper presents a starting point of the project where several traditional classification methods are compared with the originally developed classifier ensembles based on local specialization of the classifiers in given feature space partitions. The experiment was carried out on the basis of originally acquired database about patients with the different stages of liver fibrosis. The preliminary results are very promising, because they confirmed the possibility of outperforming the noninvasive commercial tests.

Liver fibrosis diagnosis support using the Dempster–Shafer theory extended for fuzzy focal elements

Abstract Classifiers are used in a variety of applications, among them the classification of medical data. Their efficiency depends on the quality of training data, which is a disadvantage in the case of medical data that are often imperfect (e.g. incomplete, imbalanced, uncertain). Moreover, numerous classifiers are black-boxes from the perspective of diagnosticians who perform the final diagnoses. These drawbacks degrade the potential usefulness of classifiers in diagnosis support. A rule-based reasoning may overcome these mentioned limitations. We introduce both a rule selection and a diagnosis support method based on the Dempster–Shafer and fuzzy set theories. The theories can manage an interpretation of incomplete and imbalanced data, imprecision of medical information and knowledge uncertainty. The usefulness of the method will be proven on a test case of liver fibrosis diagnosis. The liver fibrosis stage is difficult to recognize even for experienced physicians. The diagnosis of the liver state by an invasive biopsy is ambiguous and dependent on its finite precision. Therefore, knowledge-based methods are being sought to reduce the need of invasive testing. We use a real medical database related to patients affected by hepatitis C virus to extract knowledge. The database has missing and outlying values and patients’ diagnoses are uncertain. The proposed methods provide simple diagnostic rules that are helpful in this study of liver fibrosis and in processing deficient data. The greatest benefit and novelty of the approach is the ability to assess three stages of fibrosis in a non-invasive way, whereas other medical tests allow to detect only the last stage, i.e. the cirrhosis.

Dedicated AVR bootloader for performance improvement of prototyping process

This paper presents a new bootloader for Atmels AVR ATMega microcontroller family. Presented solution compared to Optiboot is smaller and faster, leaving programmer more flash space, and reducing time required to upload, or to update firmware of the microcontroller. Concept of bootloader made the Arduino so popular, due to removing a need for specialized programming device in the prototyping stage. However, the original idea of bootloader was to enable end-user or less qualified service personnel to update firmware of the final product/embedded device.