研究者詳細 - 井上　創造

論文 - 井上　創造

分割表示238 件中 81 - 100 件目／全件表示 >>

Towards pervasive geospatial affect perception 査読有り

Raja M., Exler A., Hemminki S., Konomi S., Sigg S., Inoue S.

GeoInformatica 22 ( 1 ) 143 - 169 2018年01月

記述言語：英語掲載種別：研究論文（学術雑誌）

© 2017, Springer Science+Business Media New York. Due to the enormous penetration of connected computing devices with diverse sensing and localization capabilities, a good fraction of an individual’s activities, locations, and social connections can be sensed and spatially pinpointed. We see significant potential to advance the field of personal activity sensing and tracking beyond its current state of simple activities, at the same time linking activities geospatially. We investigate the detection of sentiment from environmental, on-body and smartphone sensors and propose an affect map as an interface to accumulate and interpret data about emotion and mood from diverse set of sensing sources. In this paper, we first survey existing work on affect sensing and geospatial systems, before presenting a taxonomy of large-scale affect sensing. We discuss model relationships among human emotions and geo-spaces using networks, apply clustering algorithms to the networks and visualize clusters on a map considering space, time and mobility. For the recognition of emotion and mood, we report from two studies exploiting environmental and on-body sensors. Thereafter, we propose a framework for large-scale affect sensing and discuss challenges and open issues for future work.

DOI： 10.1007/s10707-017-0294-1

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85014904338&origin=inward

Deep Recurrent Neural Network for Mobile Human Activity Recognition with High Throughput 査読有り

井上聖也, 井上創造, 西田健

Journal of Artificial Life and Robotics 1 - 13 2017年12月

記述言語：英語掲載種別：研究論文（学術雑誌）

In this paper, we propose a method of human activity recognition with high throughput from raw accelerometer data applying a deep recurrent neural network (DRNN), and investigate various architectures and its combination to find the best parameter values. The “high throughput” refers to short time at a time of recognition. We investigated various parameters and architectures of the DRNN by using the training dataset of 432 trials with 6 activity classes from 7 people. The maximum recognition rate was 95.42% and 83.43% against the test data of 108 segmented trials each of which has single activity class and 18 multiple sequential trials, respectively. Here, the maximum recognition rates by traditional methods were 71.65% and 54.97% for each. In addition, the efficiency of the found parameters was evaluated using additional dataset. Further, as for throughput of the recognition per unit time, the constructed DRNN was requiring only 1.347 ms, while the best traditional method required 11.031 ms which includes 11.027 ms for feature calculation. These advantages are caused by the compact and small architecture of the constructed real time oriented DRNN.

その他リンク： https://doi.org/10.1007/s10015-017-0422-x

Recognition of Multiple Overlapping Activities Using Compositional CNN-LSTM Model 査読有り

大北剛, 井上創造

ACM Int'l Conf. Pervasive and Ubiquitous Computing (Ubicomp) Poster 2017年09月

Quality Requirements for Electronic Health Record Systems A Japanese-German Information Management Perspective 査読有り

A. Winter, K. Takabayashi, F. Jahn, Eizen Kimura, R. Engelbrecht, R. Haux, M. Honda, U. H. Hubner, 井上創造, C. D. Kohl, T. Matsumoto, Y. Matsumura, K. Miyo, 中島直樹, H. -U. Prokosch, M. Staemmler

Journal of Methods of Information in Medicine 56 92 - 104 2017年09月

記述言語：英語掲載種別：研究論文（学術雑誌）

Background: For more than 30 years, there has been close cooperation between Japanese and German scientists with regard to information systems in health care. Collaboration has been formalized by an agreement between the respective scientific associations. Following this agreement, two joint workshops took place to explore the similarities and differences of electronic health record systems (EHRS) against the background of the two national healthcare systems that share many commonalities. Objectives: To establish a framework and requirements for the quality of EHRS that may also serve as a basis for comparing different EHRS. Methods: Donabedian’s three dimensions of quality of medical care were adapted to the outcome, process, and structural quality of EHRS and their management. These quality dimensions were proposed before the first workshop of EHRS experts and enriched during the discussions. Results: The Quality Requirements Framework of EHRS (QRF-EHRS) was defined and complemented by requirements for high quality EHRS. The framework integrates three quality dimensions (outcome, process, and structural quality), three layers of information systems (processes and data, applications, and physical tools) and three dimensions of information management (strategic, tactical, and operational information management). Conclusions: Describing and comparing the quality of EHRS is in fact a multidimensional problem as given by the QRF-EHRS framework. This framework will be utilized to compare Japanese and German EHRS, notably those that were presented at the second workshop.

Kyutacar

その他リンク： https://doi.org/10.3414/ME17-05-0002

介護サービス向上に向けた介護事故事例テキストの分析(Text Analysis of Incident Reports for Preventing Accidents in Caregiving}) 査読有り

峯崎智裕, 松木萌, 井上創造

情報処理学会論文誌 58 ( 10 ) 2017年07月

記述言語：日本語掲載種別：研究論文（学術雑誌）

本論文は，複数介護施設における事故報告書を使用して，テキストマイニングおよび機械学習により事故の起こりうる状況を分析した結果について述べる．

まず，テキストデータと多変量データが混合したデータに対して，テキストマイニングおよびランダムフォレスト法を組み合わせて，事故の結果に対して影響する要因を抽出する方法を提案した．

まずテキストデータについては分かち書きを行い単語文書行列を構成し，単語についてクラスタリングを行った結果を元の多変量データと結合し，事故の結果として目的変数を設定しランダムフォレスト法￥new{を2段階で適用しながら重要な単語クラスタ及び重要な単語と属性}を抽出し，決定木によって可視化するという手法である．

この手法を，介護施設から得られた5,189件の事故事例について適用したところ，

原因不明のあざの発見のされ方や，転倒・ずり落ちから打撲につながるような行動，誤嚥につながる状況，無断外出につながる行動といった幾つかのパターンを抽出できた．

Kyutacar

センサーと医療ビッグデータを活用した医療サービス分析システムの研究開発—業務分析から課題を抽出し改善につなげる

杉山康彦, 白水麻子, 中島直樹, 井上創造

看護管理 ( 看護管理 ) 27 ( 8 ) 658 - 667 2017年06月

携帯センサと近接センサを用いた看護師業務分析査読有り

磯田達也, 井上創造, 花沢明俊, 野原康伸, 白水麻子, 杉山康彦, 平田真理, 町田京子, 中島直樹

情報処理学会論文誌 57 ( 10 ) 2197 - 2209 2017年01月

記述言語：日本語掲載種別：研究論文（学術雑誌）

本研究では,看護師の業務分析のため,33 日間に渡り看護師の業務行動データを収集する実験を行い,業務中の看護師の行動の激しさおよび,業務時間に影響を与える要因について分析を行った.データ収集実験では,38 人の看護師を被験者に設定して,装着型の小型センサ端末を装着した状態で業務を行ってもらい,加速度データおよび看護師間の対面情報,看護師業務情報を収集した.また,赤外線を用いた情報通信機器によって看護師の位置情報も同時に収集した.本稿では,看護師業務中の行動の激しさを目的変数,看護師業務毎の回数,対面回数,場所毎の訪問回数などを説明変数に設定して,決定木 (回帰木) を用いた要因分析を行った.その結果,他の看護師との対面回数が多く,病室に何度も訪れている看護師や,リハビリ介助業務や行動介助業務などの患者の介助に関わる業務を行う看護師ほど,行動の激しさが大きくなるという結果を得た.また,業務時間が業務効率に直結していると考え,各業務時間の長さに影響を及ぼす要因について回帰分析によって調べた.その後,単回帰分析により有意水準 5%で有意となった説明変数の業務だけを抽出し,その業務に対して決定木を用いた要因分析を行った.単回帰分析の結果, 14 の業務において有意な結果が得られた.また,決定木による要因分析の結果,ほとんどの業務において他の看護師との対面人数が業務時間に影響を与えていることがわかった.また,食事介助業務中に特定の場所に訪れる看護師ほど業務時間が長くなるという結果を得た.

Kyutacar

ウェアラブルセンサを用いたヒューマンセンシング

井上創造

日本知能情報ファジィ学会誌 ( 日本知能情報ファジィ学会誌 ) 170 - 186 2017年01月

Recognizing and Understanding Nursing Activities for a Whole Day with a Big Dataset 査読有り

井上創造, 上田修功, 野原康伸, 中島直樹

Journal of Information Processing 24 ( 6 ) 853 - 866 2017年01月

記述言語：英語掲載種別：研究論文（学術雑誌）

In this paper, we provide a real nursing data set for mobile activity recognition that can be used for supervised machine learning, and big data combined the patient medical records and sensors attempted for 2 years, and also propose a method for recognizing activities for a whole day utilizing prior knowledge about the activity segments in a day. Fur- thermore, we demonstrate data mining by applying our method to the bigger data with additional hospital data. In the proposed method, we 1) convert a set of segment timestamps into a prior probability of the activity segment by exploit- ing the concept of importance sampling, 2) obtain the likelihood of traditional recognition methods for each local time window within the segment range, and, 3) apply Bayesian estimation by marginalizing the conditional probability of estimating the activities for the segment samples. By evaluating with the dataset, the proposed method outperformed the traditional method without using the prior knowledge by 25.81% at maximum by balanced classification rate, and outperformed by 6.5% the F-measure with accepting 1 hour of margins. Moreover, the proposed method significantly reduces duration errors of activity segments from 324.2 seconds of the traditional method to 74.6 seconds at maximum. We also demonstrate the data mining by applying our method to bigger data in a hospital.

Kyutacar

集会報告　マルチメディア，分散，協調とモバイル（DICOMO2017）シンポジウム

井上創造, 松木萌, 岡崎有彩, 清田陽司

情報管理 ( 国立研究開発法人科学技術振興機構 ) 60 ( 8 ) 603 - 607 2017年01月

Supervised and Unsupervised Transfer Learning for Activity Recognition from Simple In-home Sensors 査読有り

井上創造, 潘新程

International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (MobiQuitous) (Acceptance Rate: 30%) 20 - 27 2016年11月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Hiroshima Hiroshima

In this paper, we propose an approach to improve the accuracy of life activity estimation using device-free sensors in the home.

We assumed two scenarios in which we only have training data from other households with labels and we also have training labels for our own household and, proposed the method to compose supervised transfer between labeled data and unsupervised for each scenario.

We developed the system which consists of an smartphone apps and web server, and gathered subjects from open called households during a period of approximately four months, and obtained approximately 11,745 activity inputs, 7.14GB of sensor data, and power consumption data of 237,280 hours from 35 households, and found to outperform naive methods in the both scenarios.

Kyutacar

その他リンク： http://dl.acm.org/citation.cfm?doid=2994374.2994400

Mobile Activity Recognition from Training Labels with Inaccurate Activity Segments (Best paper candidate) 査読有り

戸田隆道, 井上創造, 上田修功

International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (MobiQuitous) (Acceptance Rate: 30%) 57 - 64 2016年11月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Hiroshima Hiroshima

In this paper, we propose an approach to improve mobile activity recognition given training dataset with inaccurate segments, in which the beginning and the ending timestamps of homogeneous and continuous activities have inaccurate boundaries by human errors.

In the proposed approach, we A) convert the training dataset to multi-label samples, B) train the dataset with multi-label EM learning algorithm, and C) apply a segmentation method using not only the estimated labels, but also the original segment information.

We evaluate the proposed approach for 3 datasets including simulation data and real activity data, 2 machine-learning algorithms, and various inaccuracies, and show that the proposed approach outperformed the naive methods for the questions of Q1) fixing the segments of the training data and Q2) improving the recognition accuracy by cross validation.

Kyutacar

その他リンク： http://dl.acm.org/citation.cfm?doid=2994374.2994400

Activity recognition and future prediction in hospitals 査読有り

Inoue S.

ACM International Conference Proceeding Series 28-November-2016 59 - 65 2016年11月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

© 2016 ACM. In this paper, we firstly introduce our work [9] where we proposed for recognizing whole day activities using prior knowledge, and applied the method for real nursing sensor dataset we have collected. Then, we introduce the method for predicting the near future of nurses by integrating nurse activity data, location data, and medical records, based on our work [10] . For both works, we independently collected real and open nursing datasets with 2 weeks of accelerometers and training labels from 22 nurses for the former work, and nurse activity, location, medical payment, and nursing need data from 35 nurses and 96 patients for 40 days for the latter work.

DOI： 10.1145/3004010.3004012

Scopus

その他リンク： https://www.scopus.com/inward/record.uri?partnerID=HzOxMe3b&scp=85008187825&origin=inward

機械学習で怖い「次元の呪い」手法の選択は適材適所で

井上創造

まるわかり！人工知能最前線：日経コンピュータ ( まるわかり！人工知能最前線：日経コンピュータ ) 98 - 193 2016年10月

Predicting Daily Nursing Load from Nurses’ Activity Logs and Patients’ Medical Records 査読有り

井上創造, 磯田達也, 白水麻子, 杉山康彦, 野原康伸, 中島直樹

ACM Int'l Conf. Pervasive and Ubiquitous Computing (Ubicomp) Poster 4 pages 2016年09月

Recognizing Unknown Activities Using Semantic Word Vectors and Twitter Timestamps 査読有り

松木萌, 井上創造

Ubicomp Workshop for Human Activity Sensing Corpus and its Application (HASCA) 823 - 830 2016年09月

記述言語：英語掲載種別：研究論文（国際会議プロシーディングス）

Heidelberg, Germany Heidelberg, Germany

In this paper, we propose a method for activity recognition, which can estimate new activities which does not appear

in the training data, combining word vectors constructed from semantic word vectors and Twitter timestamps. Be- cause traditional activity recognition utilizes supervised machine learning, unknown activity classes which do not appear in the training dataset is unable to be estimated. For this problem, zero-shot machine learning method is pro- posed, but it requires the preparation of semantic codes.

As semantic codes, we utilize word vectors constructed from semantic word vectors and Twitter timestamps. To evaluate the proposed method, we evaluated whether we could estimate unknown activity classes, with the sensor data set collected from 20 households for 4 months, along with the user-generated labels using the web system which can estimate, modify, and add new activity types. As a re- sult, the proposed method could even estimate unknown activity classes. Moreover, by utilizing Twitter timestamps and semantic word vectors from the Japanese Wikipedia in word vectors, the method could estimate 9 unknown activity classes.

Kyutacar

その他リンク： http://hasca2016.hasc.jp

機械学習をIoTに応用人間の行動を認識・予測する

井上創造

日経コンピュータ ( 日経コンピュータ ) 78 - 81 2016年09月

機械学習をIoTに応用、人間の行動を認識・予測する

井上創造

日経ITPro ( 日経ITPro ) 2016年08月

機械学習で怖い「次元の呪い」手法の選択は適材適所で

井上創造

日経コンピュータ ( 日経コンピュータ ) 86 - 89 2016年08月

あなたの仕事を劇的に効率化！機械学習のインパクト第3回機械学習で怖い「次元の呪い」、手法の選択は適材適所で

井上創造

日経ITPro ( 日経ITPro ) 2016年08月

　前のページ - 次のページ　

<井上 創造>

論文 - 井上 創造

<井上　創造>

論文 - 井上　創造