Overfitting is a crucial problem in deep neural networks, even in the latest network architectures. In this paper, to relieve the overfitting effect of ResNet and its improvements (i.e., Wide ResNet, PyramidNet, and ResNeXt), we propose a new regularization method called ShakeDrop regularization. ShakeDrop is inspired by Shake-Shake, which is an effective regularization method, but can be applied to ResNeXt only. ShakeDrop is more effective than Shake-Shake and can be applied not only to ResNeXt but also ResNet, Wide ResNet, and PyramidNet. An important key is to achieve stability of training. Because effective regularization often causes unstable training, we introduce a training stabilizer, which is an unusual use of an existing regularizer. Through experiments under various conditions, we demonstrate the conditions under which ShakeDrop works well.

Graphic novels such as comic books and mangas are well known all over the world. The digital transition started to change the way people are reading comics: more and more on smartphones and tablets, and less and less on paper. In recent years, a wide variety of research about comics has been proposed and might change the way comics are created, distributed and read in the future. Early work focuses on low level document image analysis. Comic books are complex; they contains text, drawings, balloons, panels, onomatopoeia, etc. Different fields of computer science covered research about user interaction and content generation such as multimedia, artificial intelligence, human–computer interaction, etc. with different sets of values. We review the previous research about comics in computer science to state what has been done and give some insights about the main outlooks.

The automated generation of fonts containing a large number of characters is in high demand. For example, a typical Japanese font requires over 1,000 characters. Unfortunately, professional typographers create the majority of fonts, resulting in significant financial and time investments for font generation. The main contribution of this paper is the development of a method that automatically generates a target typographic font containing thousands of characters, from a small subset of character images in the target font. We generate characters other than the subset so that a complete font is obtained. We propose a novel font generation method with the capability to deal with various fonts, including a font composed of distinctive strokes, which are difficult for existing methods to handle. We demonstrated the proposed method by generating 2,965 characters in 47 fonts. Moreover, objective and subjective evaluations verified that the generated characters are similar to the original characters.

Mobile and wearable devices became pervasive in daily life. The dominant input techniques for mobile and wearable technology are touch and speech. Both approaches are not appropriate in all settings. Therefore, we propose a novel interface that is controlled through the tongue. It is based on an array of textile pressure sensors attached to the user's cheek. It can be easily integrated into helmets or face masks in a non-invasive way. In an initial study, we investigate gestures for tongue-based interface. Six participants repeatedly performed five simple tongue gestures. We show that gestures can be recognized with 98% accuracy. Based on feedback from participants, we discuss potential use cases and provide an outlook on further improvement of the system.