| BRIONES JERIC CRUZ | D, 中間発表 | 数理情報学 | 池田 和司, 笠原 正治, 久保 孝富, Nishanth Koganti |
|
title: Hierarchical Segmentation Approach to Detecting Switching Interaction using Bayesian Non-Parametrics
abstract: Studies on interaction detection generally assume that interaction information is time-invariant. However, interaction may change over time, especially when time series dynamics is considered. To detect these switching interaction, we propose a hierarchical segmentation-based approach to interaction detection using Bayesian non-parametrics. In today's talk, we first consider a method combining the segmentation of BP-AR-HMM and the double articulation assumption of NPYLM. Initial results of our experiments on synthesized toy and motion datasets indicate that our method is capable of segmenting motion sequences with good accuracy. The resulting AR models of each segment could then be used to detect interaction using non-parametric tests. language of the presentation: English | |||
| 松田 大樹 | M, 2回目発表 | 大規模システム管理 | 笠原 正治, 安本 慶一, 笹部 昌弘 |
|
title:Refuge Selection Scheme Considering Speedy, Reliable, and Capacity-Aware Evacuation in Mobile-Edge Collaborative Automatic Evacuation Guiding System
abstract:When a large-scale disaster occurs, evacuees have to move to refuges speedily and safely. To tackle this issue, mobile-edge collaborative automatic evacuation guiding scheme has been proposed, which combines information about road condition collected during evacuation with pre-obtained risk information. However, this scheme selects the nearest refuge without considering the refuge capacity and the route safety. In this presentation, we propose a refuge selection scheme for speedy, reliable, capacity-aware evacuation based on mobile-edge collaborative automatic evacuation guiding scheme. The proposed scheme first calculates the short and reliable routes to each refuge with the help of the existing scheme. Then, it solves integer linear programming (ILP) for refuge selection, which maximizes the route reliability under the constraints on the refuge capacity and the route length. By solving the ILP, we show that the proposed scheme can avoid the overflow of refuge capacity and improve the route reliability by 20.9% with 12.5% increase of route length, compared with the nearest refuge selection. language of the presentation:Japanese | |||
| 寺田 周平 | M, 2回目発表 | サイバネティクス・リアリティ工学 | 清川 清, 安本 慶一, 酒田 信親 |
|
title: Designing of an overeating suppression system by visual and haptic feedback on food intake
abstract: One of the values we pursue in a meal is the sense of satiety. However, continuing pursuit of satiety with daily meals may result in loss of control of food intake and health problems. Therefore, in this study, we propose an T-shirt style device that presents visual and haptic stimulus in order to reduce food intake while pursuing the sense of satiety. By sensing the amount of food intake by the user and presenting visual and haptic feedback at an appropriate timing according to the result, the system will make the user feel he or she has eaten too much when it actually is a properamount. In this way, our system improves the user's self-control and aids suppression of food intake.In this presentation, we describe an overview of the proposed system and the prototype under development. language of the presentation: Japanese | |||