We propose a new activation function and verify its performance in a deep image prior network. The new activation function is RSwish, which is Swish with a random slope for x < 0 assuming that the input value of the activation function is x. In addition, we manipulate the probability of the random number and observe the effect. As a result, we found that RSwish can perform better than Swish by manipulating probabilities according to the degree of color change in the image. © 2020 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Generative adversarial nets (GANs) perform well on a variety of tasks, but rely on large datasets and expensive computer‐based learning. SinGAN was proposed as a GANs that overcomes this problem, but its super‐resolution performance for large‐scale natural images was not very good. In this study, we aimed to improve the performance of SinGAN by changing the activation function. As a result of the verification, it was found that applying RSwish to the generator and Swish to the discriminator can generate an image with less deterioration as a whole than the default. © 2021 Institute of Electrical Engineers of Japan. Published by Wiley Periodicals LLC.

Abstract. Understanding and attention value of the advertisement will be advanced by adding scents to the digital signage. However, it was difficult to have corresponding one-to-many relationships, movements of users, and precise chronological control of scents. In this study, using mobile olfactory display, we propose the digital signage with scent which takes account of users' movements. The concept of this study is constructing the system having scents, movements, and communication. This system was built by enabling to receive the scent ejection signal from the advertisement, achieve the distance by the image of web-camera, and eject the scents with the strength in accordance with distance, to have the control of scents which accedes to substance of advertisement and positional relationship. As a result of evaluations, the olfactory information was carried down to users with great accuracy. The scents production of the advertisements will be possible with the use of this system.

No abstract

Thinking states are very important factors in the evaluation of and interaction with virtual spaces. Physiological information is often used to estimate thinking states objectively. However, it is difficult to estimate complex feelings numerically from limited physiological information. In this paper, we propose a method for the evaluation of thinking states in cyberspace using multiple physiological parameters. We developed a mapping matrix that converted physiological data into a composite of thinking states. In experiments, we found that with one mapping matrix the thinking states of different subjects for the same task could be derived. In addition, we investigated the similarities and differences between six estimated thinking states by using the mapping matrix.