| THUWIBA QAIS ABDALLA IBRAHIM | D, 中間発表 | 数理情報学 | 池田 和司 | 荒牧 英治 | 久保 孝富 | 日永田 智絵 | Li Yuzhe |
| 中川 綾子 | M, 2回目発表 | 計算行動神経科学 | 田中 沙織 | 池田 和司 | 荒牧 英治 | 川島 一朔 | |
|
title:Subtyping of Problematic Smartphone Use (PSU) by multivariable questionnaire of individual characteristics and smartphone use
abstract:Smartphones have reached 90.1% ownership among households in Japan in just 10 years since their introduction in 2013. With this rapid penetration, the Problematic Smartphone Uses (PSUs) has emerged as a new social issue. This PSUs leads to various physical and psychosocial issues, such as disrupted daily routines and impaired interpersonal relationships. However, the underlying mechanisms of PSUs remain unclear, and effective treatments are yet to be established. In this study, we conducted an online questionnaire survey on the duration and purpose of smartphone use, as well as a variety of personal characteristics. We analyzed the data collected from 583 individuals using machine learning to identify the underlying features of PSU. language of the presentation: Japanese | |||||||
| 市原 有生希 | M, 2回目発表 | 数理情報学(計算神経科学) | 池田 和司☆ | 川鍋 一晃 | 杉本 徳和 | 田中 沙織 | |
| title:Evaluation of Regularized Best-of-N Sampling Strategies for Language Model Alignment abstract: BoN (Best-of-N) sampling with a reward model helps align Large Language Models (LLMs) with human preferences but is vulnerable to reward hacking, where over-optimizing the reward model can reduce true objective performance. Previous work introduced a regularization strategy to counter this issue, showing empirical success. However, the reason behind this improvement is unclear. This study analyzes the impact of regularization on BoN sampling both empirically and analytically. language of the presentation:Japanese | |||||||
| 上原 勲紀 | M, 2回目発表 | 数理情報学(計算神経科学) | 池田 和司☆ | 川鍋 一晃 | 杉本 徳和 | 田中 沙織 | |
|
title: Collection of various demonstrations using two manipulators
abstract: Currently, laboratories around the world are collaborating to collect data sets for when robots are used to learn imitation and other tasks. However, the problem is that the types of robots used to collect the data and the tasks performed with the robots are very uneven. In this study, we will examine the possibility of collecting a wide variety of data by performing various tasks with both arms, not just operations that can be performed with one arm. We will also consider building a system to collect data by preparing an environment that reproduces a kitchen in order to increase the complexity of the environment, and to investigate the differences between teaching using a haptic device and teaching using VR. language of the presentation: Japanese | |||||||