| 鳥羽 望海 | M, 2回目発表 | ユビキタスコンピューティングシステム | 安本 慶一, 中村 哲, 諏訪 博彦, 藤本 まなと |
|
title: Comparison of Machine Learning Models in Estimating Emotions of Online Meeting Participants with Evaluations from the Administrator's Perspective
abstract: By determining the level of work engagement, which represents the psychological state of workers, supervisors and industrial physicians can grasp favorable or unfavorable conditions of workers in advance, and thereby improve the management of the management of healthy organization. On the other hand, COVID-19 has led to the adoption of remote work by workers, and it is difficult for supervisors and industrial physicians to grasp the psychological state of workers during remote work. In the present study, we focus on web conferencing services, which have the advantages of reducing costs, labor, and risk of infection, and investigate a method for estimating workers' emotions using multi-modal data such as video and audio data obtained from web conferencing. The states of participants in a group discussion using a video-telephony service are estimated by sensing various data such as video and heart rate. Using four types of indices (emotional polarity, facial landmark, heart rate, and emotional annotations) obtained from speech recordings of the participants during the group discussion and through analysis by Light Gradient Boosting Machine (LightGBM), Support Vector Regression (SVR), and our own supervisory evaluation, we found that LightGBM and SVR, which are machine learning models output better results in five out of six sessions as compared to the supervisory evaluation, concerning the mean absolute error (MAE). language of the presentation: Japanese 発表題目: オンラインミーティング参加者の感情推定における機械学習モデルと管理者から見た評価との比較 発表概要: 労働者の心理状態を表すWork Engagementを知ることで，上司や産業医は労働者の好不調を事前に把握することができ，結果的に健全な組織運営を実現することができる．一方，昨今のCOVID-19の影響で，労働者がテレワークが取り入れられるようになり，上司や産業医にとってテレワーク中の労働者の心理状態は把握しにくい．我々は費用と労力，感染リスクを減らせるメリットがあるオンラインミーティングサービスに着目し，オンラインミーティングで得られる動画や音声などのマルチモーダルデータを用いて労働者の感情を推定する手法を検討する．オンラインミーティングサービスを用いてグループディスカッションを参加者にしてもらい，そこから映像，心拍など様々なデータをセンシングして参加者の状態を推定する．グループディスカッション中の参加者の発言録から得られた感情極性，顔のランドマーク座標，心拍，他者に対する感情アノテーションの4種類の指標から，他者への感情アノテーションを推定するためにLight GBM (Gradient Boosting Machine)，SVR (Support Vector Regression)，我々自身による客観的評価を行った結果，MAEに関して機械学習モデルであるLight GBM，SVRのほうが客観的評価よりも6つのセッションのうち5つのセッションで良い結果を出力することがわかった． | |||
| 宮本 佳奈 | D, 中間発表 | 知能コミュニケーション | 中村 哲, 安本 慶一, 田中 宏季 |
|
title: Applying Meta-Learning and Iso Principle for Development of EEG-Based Emotion Induction System
abstract: Music is often used for emotion induction. Since the emotions felt when listening to it vary from person to person, customized music is required. Our previous work designed a music generation system that created personalized music based on participants’ emotions predicted from EEG data. Although our system effectively induced emotions, unfortunately, it suffered from two problems. The first is that a long EEG recording is required to train emotion prediction models. In this presentation, we trained models with a small amount of EEG data using meta-learning. The second problem is that the generated music failed to consider the participants’ emotions before they listened to music. We solved this challenge by constructing a system that adapted an iso principle that gradually changed the music from close to the participants’ emotions to the target emotion. Our results showed that emotion prediction with meta-learning had the lowest RMSE compared to other methods (p<0.016). Both a music generation system based on the iso principle and our conventional music generation system more effectively induced emotion than music generation that was not based on the emotions of the participants (p<0.016). language of the presentation: Japanese | |||