WEKO3
アイテム
強化学習を用いた行動時刻予測の通知最適解
https://doi.org/10.18997/0002000295
https://doi.org/10.18997/000200029572110864-5823-42ef-9566-c78083a68e7f
| 名前 / ファイル | ライセンス | アクション |
|---|---|---|
sei_k_469.pdf (5.5 MB)
|
|
| アイテムタイプ | 学位論文 = Thesis or Dissertation(1) | |||||||
|---|---|---|---|---|---|---|---|---|
| 公開日 | 2023-11-29 | |||||||
| 資源タイプ | ||||||||
| 資源タイプ識別子 | http://purl.org/coar/resource_type/c_db06 | |||||||
| 資源タイプ | doctoral thesis | |||||||
| タイトル | ||||||||
| タイトル | Intelligent Reminder System Using Mobile Activity Recognition | |||||||
| 言語 | en | |||||||
| タイトル | ||||||||
| タイトル | 強化学習を用いた行動時刻予測の通知最適解 | |||||||
| 言語 | ja | |||||||
| 言語 | ||||||||
| 言語 | eng | |||||||
| 著者 |
Muhammad Fikry,
× Muhammad Fikry,
|
|||||||
| 抄録 | ||||||||
| 内容記述タイプ | Abstract | |||||||
| 内容記述 | In this thesis, we propose new methods to remind people to do their daily activities, calculate the estimated time and predict the best time to remind users, and future activities, and consider whether an activity needs to be performed and whether it needs to be reminded. This is important for addressing the existing challenges related to short-term and long-term memory caused by prospective memory problems, various busyness, distractions, and multi-routine plan problems. We present methods that effectively learn the need for notification or non-notification, when, and how much time to notify by considering the balance between exploration and exploitation. Considering the balance between exploring new time options and exploiting the learned optimal timing policy can be a challenge resolved in this thesis because the reward function needs to align with the desired objectives of the reminder system, such as maximizing user responsiveness or optimizing the balance between effectiveness and user experience. Additionally, in the approaches in this thesis, we propose a notification system for forecasted activities that have not been seen in related work for forecasted activities. In forecasted activity, we need to consider whether the activity needs to be done and needs notification. We are focused on achieving low-probability increases in user activity and user engagement in responding to notifications. These proposed methods will be very helpful for the development of human activity recognition systems, particularly for people with memory problems or busy schedules, and and healthcare institution such as hospital because we can remind them well that the execution time of notifications is on target, so as to prevent users from stressing out over a lot of notifications, but those who miss notifications can receive them back at a later time step so that the activity information to be completed is still available. Although previous work has used several techniques to remind the users, the difference in our approach is that other researchers have focused on activities that have been scheduled or rescheduled to manage time, then reminded users at one time based on a schedule, whereas daily activities in the future are unpredictable and a variety of factors can make them miss reminders, so we contribute to the addition of forecasting and provide several dynamic alternative reminder times that our model will optimize. Our findings can be used by researchers and practitioners to enhance the quality and quantity of activity data collection, the accuracy of human activity recognition, and user engagement rates. |
|||||||
| 言語 | en | |||||||
| 目次 | ||||||||
| 内容記述タイプ | TableOfContents | |||||||
| 内容記述 | 第1章 Introduction|第2章 Related Works|第3章 Requirements Analysis|第4章 Reminder System Using Reinforcement Learning|第5章 Notifications for Forecasted Activity|第6章 Deploying Intelligent Technology in Healthcare Institution|第7章 Discussion and Future Work|第8章 Conclusion | |||||||
| 言語 | ja | |||||||
| 備考 | ||||||||
| 内容記述タイプ | Other | |||||||
| 内容記述 | 九州工業大学博士学位論文 学位記番号：生工博甲第469号 学位授与年月日: 令和5年9月25日 | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Reminder System | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Notification System | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Activity Recognition | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Daily Activity | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Forecasted Activity | |||||||
| キーワード | ||||||||
| 主題Scheme | Other | |||||||
| 主題 | Reinforcement Learning | |||||||
| 学位授与番号 | ||||||||
| 学位授与番号 | 甲第469号 | |||||||
| 学位名 | ||||||||
| 学位名 | 博士（工学） | |||||||
| 学位授与年月日 | ||||||||
| 学位授与年月日 | 2023-09-25 | |||||||
| 学位授与機関 | ||||||||
| 学位授与機関識別子Scheme | kakenhi | |||||||
| 学位授与機関識別子 | 17104 | |||||||
| 学位授与機関名 | 九州工業大学 | |||||||
| 学位授与年度 | ||||||||
| 内容記述タイプ | Other | |||||||
| 内容記述 | 令和5年度 | |||||||
| 出版タイプ | ||||||||
| 出版タイプ | VoR | |||||||
| 出版タイプResource | http://purl.org/coar/version/c_970fb48d4fbd8a85 | |||||||
| アクセス権 | ||||||||
| アクセス権 | open access | |||||||
| アクセス権URI | http://purl.org/coar/access_right/c_abf2 | |||||||
| ID登録 | ||||||||
| ID登録 | 10.18997/0002000295 | |||||||
| ID登録タイプ | JaLC | |||||||
