Munehiro Nakamura

LVQ-SMOTE - Learning Vector Quantization based Synthetic Minority Over-sampling Technique for biomedical data.

BackgroundOver-sampling methods based on Synthetic Minority Over-sampling Technique (SMOTE) have been proposed for classification problems of imbalanced biomedical data. However, the existing over-sampling methods achieve slightly better or sometimes worse result than the simplest SMOTE. In order to improve the effectiveness of SMOTE, this paper presents a novel over-sampling method using codebooks obtained by the learning vector quantization. In general, even when an existing SMOTE applied to a biomedical dataset, its empty feature space is still so huge that most classification algorithms would not perform well on estimating borderlines between classes. To tackle this problem, our over-sampling method generates synthetic samples which occupy more feature space than the other SMOTE algorithms. Briefly saying, our over-sampling method enables to generate useful synthetic samples by referring to actual samples taken from real-world datasets.ResultsExperiments on eight real-world imbalanced datasets demonstrate that our proposed over-sampling method performs better than the simplest SMOTE on four of five standard classification algorithms. Moreover, it is seen that the performance of our method increases if the latest SMOTE called MWMOTE is used in our algorithm. Experiments on datasets for β-turn types prediction show some important patterns that have not been seen in previous analyses.ConclusionsThe proposed over-sampling method generates useful synthetic samples for the classification of imbalanced biomedical data. Besides, the proposed over-sampling method is basically compatible with basic classification algorithms and the existing over-sampling methods.

Quantitative evaluation of pneumoconiosis in chest radiographs obtained with a CCD scanner

This paper presents a computer-aided diagnosis for pneumoconiosis radiographs obtained with a common CCD scanner. Since the existing diagnosis systems for pneumoconiosis extract abnormalities of pneumoconiosis from images obtained with a special scanner which can appropriately apply for chest radiographs, it is difficult to apply the methods to images obtained with a CCD scanner due to unclear shadow and the systems are not practical for medical doctors due to high costs. In the unclear images, the abnormal levels of pneumoconiosis could depend on density distributions in each of intercostal and rib areas. Therefore, the proposed method measures the abnormalities by extracting characteristics of the distribution in the areas. Besides, the proposed method classifies the images into the three categories of pneumoconiosis. Experimental results of the classifications for 51 right-lung images including 6 pneumoconiosis images have shown that the proposed abnormalities are well extracted according to the standards of pneumoconiosis categories.

A link selection method for web‐browser using eye‐gaze input

Methods for browsing Web pages using eye-gaze input have been proposed for severely physically handicapped people who cannot handle a computer mouse to utilize convenient services on the Internet. There are two important functions for Web-browsers, namely scroll and link selection. Since it is difficult to introduce existing link selection methods using eye-gaze input for home usage because of high costs or complexity, we propose in this paper a novel link selection method using eye-gaze input. The proposed method analyzes the HTML source in a Web page and makes a group of links by the content of each. In evaluation experiments, we compare the proposed method with a comparative method using eye-gaze input with respect to the average response time in selecting the links. The results of the experiments show that the average response is nearly twice as fast as that of the compared method.