Search Results 1-20 of 115

  • Recognition and Translation of Japanese-English Code-switching Speech for Monolingual Speakers  [in Japanese]

    中山 佐保子 , サクティ サクリアニ , 中村 哲

    会話の中で複数の言語が切り替わる現象は,コードスイッチングと呼ばれる.コードスイッチングは,言語が切り替わる場所や長さによってさまざまなものがある.従来の音声認識システムは,そのようなコードスイッチングを扱うのが難しく,解決すべき課題の1つであった.これまで研究されてきたコードスイッチング音声認識は,言語が混ざったコードスイッチング音声を,そのまま言語が混ざったコードスイッチングテキストに書き起こ …

    情報処理学会論文誌 62(3), 903-914, 2021-03-15

    IPSJ 

  • Fast and Accurate Driver Action Recognition with Multi-Task Learning of Driver Pose and Action  [in Japanese]

    Nishiyuki Kenta , Hyuga Tadashi , Tasaki Hiroshi , Kinoshita Koichi , Hasegawa Yuki , Yamashita Takayoshi , Fujiyoshi Hironobu

    … We train our network model with multi-task learning includes localizing and detecting each body part of the driver, classifying state of each body part, and recognizing driver action at once. … Our multi-task learning for the proposed model achieves a significant improvement compared to state-of-the-art human action recognition methods with limited computational resources. …

    Transactions of the Japanese Society for Artificial Intelligence 36(2), A-K93_1-10, 2021

    J-STAGE 

  • Multi-Category Image Super-Resolution with Convolutional Neural Network and Multi-Task Learning

    URAZOE Kazuya , KUROKI Nobutaka , KATO Yu , OHTANI Shinya , HIROSE Tetsuya , NUMA Masahiro

    … <p>This paper presents an image super-resolution technique using a convolutional neural network (CNN) and multi-task learning for multiple image categories. … There are two possible solutions to manage multi-categories with conventional CNNs. … This architecture can simultaneously learn the high-resolution image and its category using multi-task learning. …

    IEICE Transactions on Information and Systems E104.D(1), 183-193, 2021

    IR  J-STAGE 

  • Dynamically Updating Event Representations for Temporal Relation Classification with Multi-category Learning

    Fei Cheng , Masayuki Asahara , Ichiro Kobayashi , Sadao Kurohashi

    … Kyoto UniversityNational Institute for Japanese Language and LinguisticsOchanomizu UniversityKyoto UniversityTemporal relation classification is a pair-wise task for identifying the relation of a temporal link (TLINK) between two mentions, i.e. event, time and document creation time (DCT). … Our model deals with three TLINK categories with multi-task learning to leverage the full size of data. …

    Findings of the Association for Computational Linguistics: EMNLP 2020, 1352-1357, 2020-11

    IR 

  • Harnessing Generated Instance-Specific Neutral Expressions in Facial Expression Classification

    BAI Wenjun , Quan Changqin , Luo Zhi-Wei

    システム制御情報学会研究発表講演会講演論文集 64, 520-526, 2020-05-20

  • End-to-End Speech Translation With Transcoding by Multi-Task Learning for Distant Language Pairs

    Kano Takatomo , Sakti Sakriani Watiasri , Nakamura Satoshi

    … To guide the attention-based encoder-decoder model on this difficult problem, we construct end-to-end speech translation with transcoding and utilize curriculum learning (CL) strategies that gradually train the network for end-to-end speech translation tasks by adapting the decoder or encoder parts. …

    2020-04-20

    IR 

  • Towards Explainable Melanoma Diagnosis: Prediction of Clinical Indicators Using Semi-supervised Learning  [in Japanese]

    村林 誠也

    … Our proposal effectively utilizes virtual adversarial training as a semi-supervised learning framework with multi-task learning. …

    法政大学大学院紀要. 理工学・工学研究科編 (61), 1-2, 2020-03-24

    IR  DOI 

  • Multi-Task Curriculum Learning for Open-Set Semi-Supervised Recognition (ITS : Intelligent Transport Systems Technology)

    郁 青 , 入江 豪 , 相澤 清晴

    電子情報通信学会技術研究報告 = IEICE technical report : 信学技報 119(421), 323-328, 2020-02-27

  • Multi-Task Curriculum Learning for Open-Set Semi-Supervised Recognition

    郁 青 , 入江 豪 , 相澤 清晴

    映像情報メディア学会技術報告 = ITE technical report 44(6), 323-328, 2020-02

  • Reduction of Learning Time for Differential-Wheeled Mobile Robots by Knowledge Transfer for Real-Time Learning  [in Japanese]

    Kotani Naoki

    Transactions of the Institute of Systems, Control and Information Engineers 33(12), 317-319, 2020

    J-STAGE 

  • Semi-Supervised Extractive Question Summarizer Using Question-Answer Pairs and its Learning Methods  [in Japanese]

    ISHIGAKI Tatsuya , Machida Kazuya , Kobayashi Hayato , TAKAMURA Hiroya , OKUMURA Manabu

    <p>本稿は質問を対象とした抽出型要約を扱う.ニューラルネットワークによる抽出型要約モデルの学習には,大規模なラベル付きデータが必要となる.ユーザが自由に記述する Yahoo! 知恵袋などのコミュニティ QA (CQA) に投稿される質問に対しては,ラベル付きデータの獲得が難しい.そこで,本研究ではラベル付きデータが不足する問題を軽減するため,小規模な人手ラベル付きデータに加え,CQA …

    Journal of Natural Language Processing 27(4), 825-852, 2020

    J-STAGE 

  • Automatically Knowledge Reuse for Regression Problems in Genetic Programming  [in Japanese]

    Shinji Kato , Nagao Tomoharu

    … GP with transfer learning has been proposed as methods of using knowledge. … Another approach is to use knowledge by multi-task learning, but it is necessary to solve multiple problems at the same time. … This method uses an island model to extract knowledge, and a machine learning model to select knowledge. …

    Transaction of the Japanese Society for Evolutionary Computation 11(3), 45-54, 2020

    J-STAGE 

  • FR-MTL: Traffic Flow Prediction Using Fused Ridge Denoising and Multi-Task Learning

    Yang Di , Qiu Ningjia , Wang Peng , Yang Huamin

    … In this work, we propose a Fused Ridge multi-task learning (FR-MTL) model for multi-road traffic flow prediction. … In addition, we jointly consider multi-task learning (MTL) for training shared spatiotemporal information among traffic roads. …

    Journal of Advanced Computational Intelligence and Intelligent Informatics 24(7), 829-836, 2020

    J-STAGE 

  • DNN-Based Full-Band Speech Synthesis Using GMM Approximation of Spectral Envelope

    KOGUCHI Junya , TAKAMICHI Shinnosuke , MORISE Masanori , SARUWATARI Hiroshi , SAGAYAMA Shigeki

    … Furthermore, we propose a method for multi-task learning based on minimizing these errors simultaneously. …

    IEICE Transactions on Information and Systems E103.D(12), 2673-2681, 2020

    J-STAGE 

  • An Experimental Consideration for Algorithm of Tracking for Resident-Tracking Robot System:-Ability of Optimize for Tracking Method-  [in Japanese]

    SUGIMOTO Masashi , TSUZUKI Shinji , HIRANO Masatsugu , YOSHIOKA Takashi , NONAKA Shogo

    … In the algorithm, a resident-tracking task and an energy-saving task are holding. … The whole tasks are working based on Reinforcement Learning. … In the algorithm, a resident-tracking task and an energy-saving task are holding.</p> …

    The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) 2020(0), 2P1-K01, 2020

    J-STAGE 

  • Multi-Task Convolutional Neural Network for Bridge Damage Assessment  [in Japanese]

    OZEKI Makoto , HORITA Shuhei , YONAHA Makoto , YAMAGUCHI Kohei , NAKAMURA Shozo

    <p>橋梁定期点検における点検診断員の不足や維持管理コストの増加に対処するために,AI技術を活用した点検の効率化が進められている.現状の点検業務の効率化を支援するためには,点検マニュアルが対象としている様々な損傷の種類・評価基準に対応した汎用的な AIモデルが求められる.本研究では,汎用的かつ高精度な損傷程度評価モデルを構築するために,損傷程度評価と関連性が高い損傷分類とのマルチタスク …

    Intelligence, Informatics and Infrastructure 1(J1), 86-91, 2020

    J-STAGE 

  • Joint Multi-Patch and Multi-Task CNNs for Robust Face Recognition

    LIU Yanfei , CHEN Junhua , QIU Yu

    … <p>In this paper, we present a joint multi-patch and multi-task convolutional neural networks (JMM-CNNs) framework to learn more descriptive and robust face representation for face recognition. …

    IEICE Transactions on Information and Systems E103.D(10), 2178-2187, 2020

    J-STAGE 

  • Personal Semantic Variations in Word Meanings: Induction, Application, and Analysis

    Oba Daisuke , Sato Shoetsu , Akasaki Satoshi , Yoshinaga Naoki , Toyoda Masashi

    task, distinguishing each word used by different individuals. … Review-target identification was adopted as a task to prevent irrelevant biases from contaminating word embeddings. … The scalability and stability of inducing personalized word embeddings were improved using a residual network and independent fine-tuning for each individual through multi-task learning along with target-attribute predictions. …

    Journal of Natural Language Processing 27(2), 467-490, 2020

    J-STAGE 

  • A Double Adversarial Network Model for Multi-Domain and Multi-Task Chinese Named Entity Recognition

    HU Yun , ZHENG Changwen

    … In some domains that small annotated training data is available, multi-domain or multi-task learning methods are often used. … In this paper, we explore the methods that use news domain and Chinese Word Segmentation (CWS) task to improve the performance of Chinese named entity recognition in weibo domain. …

    IEICE Transactions on Information and Systems E103.D(7), 1744-1752, 2020

    J-STAGE 

  • Mathematical Representation of Emotion Using Multimodal Deep Neural Networks:Effects of the Number of Dimensions of Emotional Space on the Performance of Recognition and Unification Tasks  [in Japanese]

    HARATA Seiichi , SAKUMA Takuto , KATO Shohei

    <p>ロボット内で感情を再現するためには,感情コンピューティングの各要素において,感情を計算機内で数理的に表現するモデルが必要と考える.単一のモダリティからDNNを学習し,感情を連続値のベクトルで表現する手法では,得られる感情の表現(感情空間)はそのモダリティに依存すると推察され,並びに感情空間の次元数について考察する必要があると考える.本研究では,感情空間を獲得する複数のモダリティを …

    Proceedings of the Annual Conference of JSAI JSAI2020(0), 4F2OS25a02-4F2OS25a02, 2020

    J-STAGE 

Page Top