著者
西銘 大喜 遠藤 聡志 當間 愛晃 山田 孝治 赤嶺 有平
出版者
一般社団法人 人工知能学会
雑誌
人工知能学会論文誌 (ISSN:13460714)
巻号頁・発行日
vol.32, no.5, pp.F-H34_1-8, 2017-09-01 (Released:2017-09-01)
参考文献数
20

Facial expressions play an important role in communication as much as words. In facial expression recognition by human, it is difficult to uniquely judge, because facial expression has the sway of recognition by individual difference and subjective recognition. Therefore, it is difficult to evaluate the reliability of the result from recognition accuracy alone, and the analysis for explaining the result and feature learned by Convolutional Neural Networks (CNN) will be considered important. In this study, we carried out the facial expression recognition from facial expression images using CNN. In addition, we analysed CNN for understanding learned features and prediction results. Emotions we focused on are “happiness”, “sadness”, “surprise”, “anger”, “disgust”, “fear” and “neutral”. As a result, using 32286 facial expression images, have obtained an emotion recognition score of about 57%; for two emotions (Happiness, Surprise) the recognition score exceeded 70%, but Anger and Fear was less than 50%. In the analysis of CNN, we focused on the learning process, input and intermediate layer. Analysis of the learning progress confirmed that increased data can be recognised in the following order “happiness”, “surprise”, “neutral”, “anger”, “disgust”, “sadness” and “fear”. From the analysis result of the input and intermediate layer, we confirmed that the feature of the eyes and mouth strongly influence the facial expression recognition, and intermediate layer neurons had active patterns corresponding to facial expressions, and also these activate patterns do not respond to partial features of facial expressions. From these results, we concluded that CNN has learned the partial features of eyes and mouth from input, and recognise the facial expression using hidden layer units having the area corresponding to each facial expression.
著者
玉城翔 當間愛晃 赤嶺有平 山田孝治 遠藤聡志
雑誌
第77回全国大会講演論文集
巻号頁・発行日
vol.2015, no.1, pp.47-49, 2015-03-17

入力データからの特徴抽出器としての機能を持つニューラルネットにおいて、深い階層構造の構築は、より抽象度の高い特徴表現の獲得を可能にしている。更に、この特徴の汎化能力の向上にDropoutという技術が大きく貢献している。このDropoutにおいて経験的観点でのパラメータ設定が通例だが、その理由や妥当性については十分な検証がされていない。パラメータ設定によっては学習コストが高くなることも想定されるが、問題の複雑さ、用意したニューロン数、接続前後のニューロン状況等に応じて適切なDropout率があると考えられる。そこで、我々はニューラルネットにおける評価関数の値を使い、最適なDropout率の設定が可能かどうかの検証をする。
著者
西銘 大喜 遠藤 聡志 當間 愛晃 山田 孝治 赤嶺 有平
出版者
一般社団法人 人工知能学会
雑誌
人工知能学会論文誌
巻号頁・発行日
vol.32, no.5, pp.F-H34_1-8, 2017

<p>Facial expressions play an important role in communication as much as words. In facial expression recognition by human, it is difficult to uniquely judge, because facial expression has the sway of recognition by individual difference and subjective recognition. Therefore, it is difficult to evaluate the reliability of the result from recognition accuracy alone, and the analysis for explaining the result and feature learned by Convolutional Neural Networks (CNN) will be considered important. In this study, we carried out the facial expression recognition from facial expression images using CNN. In addition, we analysed CNN for understanding learned features and prediction results. Emotions we focused on are "happiness", "sadness", "surprise", "anger", "disgust", "fear" and "neutral". As a result, using 32286 facial expression images, have obtained an emotion recognition score of about 57%; for two emotions (Happiness, Surprise) the recognition score exceeded 70%, but Anger and Fear was less than 50%. In the analysis of CNN, we focused on the learning process, input and intermediate layer. Analysis of the learning progress confirmed that increased data can be recognised in the following order "happiness", "surprise", "neutral", "anger", "disgust", "sadness" and "fear". From the analysis result of the input and intermediate layer, we confirmed that the feature of the eyes and mouth strongly influence the facial expression recognition, and intermediate layer neurons had active patterns corresponding to facial expressions, and also these activate patterns do not respond to partial features of facial expressions. From these results, we concluded that CNN has learned the partial features of eyes and mouth from input, and recognise the facial expression using hidden layer units having the area corresponding to each facial expression.</p>
著者
赤嶺 有平 遠藤 聡志 上原 和樹 根路銘 もえ子
雑誌
情報処理学会論文誌 (ISSN:18827764)
巻号頁・発行日
vol.55, no.1, pp.438-447, 2014-01-15

交通渋滞は,経済損失を発生させるだけでなく環境へ悪影響も与えるため,その解決が強く求められている.一方,スマートフォン等の普及により位置情報の取得と通信による情報共有が安価に実現できるようになっており,得られた交通情報を活用した渋滞解消策が望まれる.本論文は,プローブカー等により所要時間のリアルタイムデータの推定や過去の蓄積データが利用可能な状況下において,適切な出発時刻および経路をユーザに提示することで,時間・空間的に交通量を分散する手法を提案する.さらに,パーソントリップ調査に基づく実データを用いたシミュレーション実験によりその効果を検証する.Traffic congestion is a major problem in many modern cities because it causes large economic losses and negatively affects to the city environment. In the meantime, traffic information has been easily collectable in real time with popularization of mobile devices that are able to communicate and localize itself. A solution using the traffic information is desired. In this paper, we propose a method to spread traffic demand temporally with indication of appropriate departure time and route for user under the situation that hourly trip time is available in real time by probe cars. In addition, we prove the efficiency of the method by the traffic simulation using actual data of person trip census.
著者
伊志嶺拓人 赤嶺 有平 遠藤 聡志
雑誌
情報処理学会論文誌 (ISSN:18827764)
巻号頁・発行日
vol.51, no.10, pp.1986-1994, 2010-10-15

地方都市では,自家用車が主要な交通手段となっており,それによる交通渋滞・環境汚染が深刻化している.そのため,時差通勤やロードプライシングなど渋滞緩和,公共交通利用を促す施策が検討されているが,利用者への負担や社会的コストが大きく,普及するに至っていない.そこで本研究では,低コストで導入可能なモーダルシフト政策として,昼間時に遊休化している通勤車両を共同利用する新しい交通システム「通勤車利用型カーシェアリング(CSCC)」を提案する.本論文では,マルチモーダル交通シミュレータを用いて提案交通システムによるモーダルシフト効果の検証結果について述べる.交通シミュレータの現況再現性評価では,自家用車のみの経路配分,交通手段選択を含んだ経路配分のシミュレーションを行い,高い相関が得られた.CSCCのモーダルシフト効果のシミュレーション分析では,沖縄県那覇市と周辺地域においてモノレールと連携したCSCCの車両提供,利用需要を算出し実現可能性を検討した.次に,モノレール利用者数の算出を行い,モノレール利用の増加,自動車利用の減少が確認でき,CSCCによってモーダルシフト効果が期待できることを示した.In provincial city, people use private cars as the primary mode of transportation. it causes traffic congestion and environmental pollution. Staggered working hours and road pricing has been suggested to promote using the public transportation as the countermeasures. however, it has not spread since a user's burden and social cost are large. In this study, we propose a new transport system "Car Sharing System Using Commuter's Car" that share the commuter's car which is not used at the daytime as modal shift policy this paper describes the verification result of the modal shift effect by a proposal traffic system using multimodal traffic simulator. High reproducibility of traffic conditions has been resulted from both simulations that runs under only car distribution and distribution of various type of transportation. As preliminary experiment of simulation analysis of the modal shift effect by CSCC, we discuss feasibility study of CSCC cooperating with a monorail in the Naha city, Okinawa and the surrounding area by calcuclation of demand to provide commuter's car and to use car sharing. Finally, we estimated that the modal shift effect by CSCC is expectable by calculating monorail ridership using the simulator. The result shows that monorail user increased and private car user decreased.