Yusuke Kajiwara

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.

Estimating Distracted Pedestrian from Deviated Walking Considering Consumption of Working Memory

This paper proposes a method to distinguish distracted pedestrians from normal pedestrians, using the acceleration and the angular velocity while walking. This method uses an acceleration sensor attached on the back of the pedestrian. The acceleration and the angular velocity are obtained while the pedestrian is walking. In addition to that, walking features are calculated based on the obtained data. Some studies points out distraction of the pedestrian relates to consumption of working memory. We assume considering the relationship between consumption of working memory and walking behavior suggest the effectiveness to estimate distraction of the pedestrian. When each pedestrian is walking while consuming working memory, for example thinking about something, their walk deviates from normal. Machine Learning method, Random Forest, is applied to classify whether the pedestrian is distracted using features of walking. An experiment suggests we can distinguish walking features represent both distracted state and normal state completely. The result indicates the method can find distracted pedestrians whose working memory is highly consumed. We discuss why we can distinguish the distraction of the pedestrian from walking feature components with the variable importance. In addition, we have conducted regression analysis on the significant feature components to figure out the reasons. Finally, we discuss the feasibility of our proposed method.

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.

Guidance of Farming Works to Improve Efficiency Considering Physical Behavior

New farmers need technical guidance to improve working efficiency because they are lacking in experience. Agricultural experts put much effort to provide guidance for beginner farmers. However, continuing to give guidance is difficult because it is a large burden on the experts. This study proposes a system which contributes to transferring a deft motion of experts to improve the working efficiency of beginners in farm works. The system promotes beginners to assess their own farming works without an expert. The beginners can confirm whether their own works are proper works. An experiment has suggested that machine learning can achieve judgement of the properness of farming works using state transition probability of each body part.