INOUE, Masashi (M. INOUE)

icmi(2016/Nov) koho(2020/Oct)

Larger photos are not available.

Contact Information

ADDRESS:
Tohoku Insitute of Technology
35-1, Kasumi-cho, Yagiyama, Taihaku-ku,. Sendai, Japan, 982-8577
DIVISION:
Department of Information and Communication Engineering
POSITION:
Associate Professor
LAB:
Interaction Engineering
EMAIL:
m.inoue(a)acm.org
ORCID iD iconorcid.org/0000-0002-9364-3114

Index - [Profile] [Research Projects] [Journal Papers] [Conference Papers] [Presentations] [Other Documents]
NOTE: Please download official PDFs from publishers when possible.

Research Interests:

My research interest lies in the understanding human cognition and communication from information processing perspective. The strangeness is that we can get along with the insufficiency of information. I approach this problem by developing methodologies to construct models that can overcome information scarcity in intelligent applications. Some keywords are listed below:

Short Biography

He received the B.A. degree from International Christian University (ICU), Japan in 1999, the M.S. and D.S. degree from Nara Institute of Science and Technology (NAIST), Japan in 2001 and 2004, respectively. He was a research associate at the National Institute of Informatics (NII) in Japan from 2004 to 2009, assistant professor at Yamagata University in Japan from 2009 to 2017, and currently with Tohoku Institute of Technology.
Social Profiles: LinkedIn, ResearchGATE, Semantic Scholar, and Google Citation.
My X(Twitter) account is ENInoue.

Laboratory

My teaching/supervisory activities and lab member/visitor information are in the Interaction Engineering Laboatory page.

Research Projects:

Hyperlinks in square brackets are links to the bibliographical information within this page.

Journal Papers:

    [SENSORS21]
    Inoue, M.; Irino, T.; Furuyama, N.; Hanada, R.
    "Observational and Accelerometer Analysis of Head Movement Patterns in Psychotherapeutic Dialogue"
    Sensors, Vol.21 No.9, Article No. 3162, May. 2021
    [@MDPI]Open Access
    - This paper studies the patterns of head nodding during psychological face-to-face counseling.
    Non-verbal patterns were revealed based on the nodding frequencies counted manually by human assesors and head movement correlation obtained by IMUs (accelerometers).


    [MTAP12]
    INOUE, Masashi; OGIRHARA, Mitsunori; HANADA, Ryoko; FURUYAMA, Nobuhiro
    "Gestural cue analysis in automated semantic miscommunication annotation"
    Multimedia Tools and Applications, Vol.61 No.1, pp.7-20, Nov. 2012
    [@Springer] [pdf (author-version)] [Google Scholar Citation]
    - This paper studies the possibility to predict miscommunication from gestural infromation in dialogues.
    Several types of hand movements were extracted and machine learning based classifiers were applied to predict occurrence of miscommunication during psychotherapeutic interviews.

    [PI09]
    INOUE, Masashi
    "Image retrieval: Research and use in the information explosion"
    Progress in Informatics, No. 6, pp.3-14, March 2009.
    [abstract] [PDF] [bibtex] [Google Scholar Citation]
    - This paper reviews two different disciplines with regard to the visual information access; namely, library science and information retrieval.
    Future direction to practically improve the visual information access by combining concepts in two discipline is discussed.
    The main idea is to annotate images; then, the images are accessed through the textual annotation.


    [VLSI04]
    UEDA, Naonori and INOUE, Masashi
    "Extended Tied-Mixture HMMs for both Labeled and Unlabeled Time Series Data"
    Journal of VLSI Signal Processing. Vol. 37, Issue. 2, pp. 189-197, June/July 2004.
    --- Now called Journal of Signal Processing Systems
    Please see [PAMI03] which is more up-to-date.
    [@Springer] [Google Scholar Citation]
    --- [NNSP2001] invited for publication.

    [PAMI03]copyright notice
    INOUE, Masashi and UEDA, Naonori
    "Exploitation of Unlabeled Sequences in Hidden Markov Models"
    IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. 25, No. 12, pp. 1570-1581, December 2003.
    [@IEEE-CS] [@IEEE Xplore] [pdf] [Google Scholar Citation]

    The term "unlabeled data" is not uniquely used in the literatures concerning semi-supervised HMM. This paper calls attention to the existence of three definitions of "unlabeled sequence" in HMM and introduces a semi-supervised version of hidden Markov model architecture. A learning algorithm for one of the three definitions of unlabeled sequences is derived.
    Here is a supplemental visual aid [pdf] to clarify the three kinds of unlabeled sequences. Also, the difference between the mixture structure used in this work and a mixture of expert (MoE) structure is discussed in the first appendix(pdf) of my PhD Thesis.

Papers in conference proceedings:

    [AIRS18a]
    Masashi Inoue and Ryu Yasuhara
    Considering Conversation Scenes in Movie Summarization
    Asia Information Retrieval Symposium, 2018
    [@Springer] [author-version]

    - This paper proposed a method to improve automatic video summarization by considering continuity of conversation scenes. The summarization is conducted using both visual and auditorial features. The evaluation is based on the informativeness and interestingness that were scored by crowd assessors. The dataset used for the experiment was from the Movie Dialogue Corpus [ICMI16]

    [AIRS18b]
    Masashi Inoue and Hiroshi Ueno
    Influence of Data-Derived Individualities on Persuasive Recommendation
    Asia Information Retrieval Symposium, 2018
    [@Springer] [author-version]

    - This paper present a method to add personal style to dialogue systems focusing on the back-channeling.
    Without changing original utterances generated by the system, corpus-driven personal traits were added to the system.
    The influence of the personalities were tested in the restraunt recommendation setting where the system persuade the users to modify their choices.

    [ICMI16]
    Ryu Yasuhara, Masashi Inoue, Ikuya Suga, and Tetsuo Kosaka
    "Large-Scale Multimodal Movie Dialogue Corpus"(demo)
    18th ACM International Conference on Multimodal Interaction (ICMI2016). 12-16 Nov. 2016, Tokyo Japan
    pp. 414-415
    [@ACM-DL] [updated author-version(pdf)] Movie Dialogue Corpus

    - This paper demonstrates a dialogue corpus taken from public domain movies. There is a support page for the corpus that contain links to the data ans script and explanations. The corpus has been updated since publication. Please see the updated author-version PDF above.
    - There is an error in official paper. The parameter used for removal of speech segment was 5 seconds, not 4 seconds. The correction is included in the above update version.
    [LREC16]
    Masashi Inoue and Hiroshi Ueno
    "Dialogue System Characterization by Back-channelling Patterns Extracted from Dialogue Corpus"
    10th edition of the Language Resources and Evaluation Conference (LREC2016). 23-28 May 2016, Portoroz (Slovenia)
    pp. 2736-2740
    [summary (pdf, bibtex)] --- This presentation was supported by "The Murata Science Foundation".


    - This paper proposed a method to add characteristics to text-based dialogue systems by using back-channeling patterns. The patterns were extracted from a corpus. The back-channeling actions occurred real-time while users are typing. The use of back-channeling did not change the message content but altered perceived individualities.
    [IWEC14]
    INOUE, Masashi; UENO, Hiroshi
    "Wizard-of-Oz Support Using a Portable Dialogue Corpus"
    The 5th International Workshop on Empathic Computing. 1 Dec. 2014, Gold Coast, Australia.
    [author-version(pdf)] Post-conference prceedings paper has been published as an LNCS paper [@Springer].
    Principles and Practice of Multi-Agent Systems
    Volume 9935 of the series Lecture Notes in Computer Science pp 119-128, 2016


    - This paper proposed a dialogue data collection method using Wizard-of-Oz (WOZ) method. The dialogue data between human and dialogue systems are scarese. The proposed method utilizes existing human-to-human dialogue corpus to assist wizards in data collection process. The corpus was transformed into a portable or reusable form.
    [MMCORPORA14]
    INOUE, Masashi; IRINO, Toshio; HANADA, Ryoko; FURUYAMA, Nobuhiro; MASSAKI, Hiroyasu
    "Continuous Annotations for Dialogue Status and Their Change Points"
    International Workshop Series on Multimodal Corpora, Multimodal Corpora: Combining applied and basic research targets. pp. 1-4, 27 May 2014, Reykjavik, Iceland.
    [(pdf)]

    - This paper investigated utility of our continuous emotion annotation tool for multimodal dialogue. The annotated emotional/situational data were used to find semantic change points in a goal-oriented dialogue.
    [LREC14]
    TAKAHASHI, Kodai and INOUE, Masashi
    "Multimodal dialogue segmentation with gesture post-processing"
    The 9th edition of the Language Resources and Evaluation Conference (LREC). 26-31 May 2014, Reykjavik, Iceland.
    [author-version(pdf)] [Official summary/pdf, bibtex] [Google Scholar Citation]

    - This paper proposed a method to refine text-based dialogue segmentation method using gesture information. The presence of gestures around topic boundaries were used in the form of explicit filtering rules.
    [ITIS14]
    INOUE, Masashi and SAKUMA, Sakie
    "Analysis of the film selection process for a film festival"
    The 7th International Workshop on Information Technology for Innovative Services (ITIS-2014). 13-16 May 2014, Victoria, Canada.
    [author-version(pdf)]
    IEEE Xplore

    - This paper analyzed metadata of films rather than film contents in a film festival (YIDFF: Yamagata International Documentary Film Festival) to qualitatively understand charactrization factors of the film selection process.
    Corrigendum 1: The descriptoin "YIDFF has been held ... biannually." is wrong. It is held biennially.
    Corrigendum 2: The citation [2] has not been mentioned in the paper by mistake.
    [iCast-Umedia13]
    ENDO, Jumpei and INOUE, Masashi
    "Speed Dependencies of Human Gesture Recognition"
    International Conference on Awareness Science and Technology (iCAST2013), and the International Conference on Ubiquitous Multimedia (UMEDIA2013), Aizu-Wakamatsu, Japan, November 2-4, 2013
    [@IEEE-Xplore] [author-version(pdf)]

    - This paper present a method to experimentally analyze selected gestures by using motion-captured videos taken by Kinect sensors.
    [AAAI12Fall]
    INOUE, Masashi
    "Human Judgment on Humor Expressions in a Community-based Question-answering Service"
    Fall 2012 Symposium on Artificial Intelligence of Humor, Washington, DC, USA, November 2-4.
    [official abstract, pdf]

    - This paper forcuses on the perception of humor found in online community.
    Both degree of humorousness and humor categories were considered.
    [MMCORPORA12]
    INOUE, Masashi; HANADA, Ryoko; FURUYAMA, Nobuhiro; IRINO, Toshio; ICHINOMIYA, Takako; MASSAKI, Hiroyasu
    "Multimodal corpus for psychotherapeutic situation"
    International Workshop Series on Multimodal Corpora, Tools and Resrouces. pp.18-21, 22 May 2012, Istanbul, Turkey.
    [author-version(pdf)] [Entire official proceedings(pdf)] [Google Scholar Citation]

    - This paper explains the procedure to construct multimodal psychotherapeutic interview corpus.
    Special forcus is on the possibility to share corpus information while preserving privacies of the recorded participants.
    [LREC12]
    INOUE, Masashi and AKAGI, Toshiki
    "Collecting Humorous Expressions from a Community-based Question-answering-service Corpus"
    The eighth international conference on Language Resources and Evaluation (LREC). 21-27 May 2012, Istanbul, Turkey.
    [author-version(pdf)] [Official summary/pdf, bibtex] [Google Scholar Citation]

    - This paper proposes a semi-automatic method to collect humor expressions from community-based question-answering (CQA) corpus.
    Dissimilarity between serious answers and humorous answeres was utilized and expressions were collected efficiently.
    [HCII11a]
    INOUE, Masashi; IRINO, Toshio; FURUYAMA, Nobuhiro; HANADA, Ryoko; ICHINOMIYA, Takako; MASSAKI, Hiroyasu
    "Manual and accelerometer analysis of head nodding patterns in goal-oriented dialogues"
    Human-Computer Interaction, Part II, HCII 2011, Orlando, FL, 12 July 2011.
    LNCS 6762, pp.259-267, Springer
    [@Springer] [author-version(pdf)] [Google Scholar Citation]

    Please see [SENSORS21] that is updated work on this topic.
    [HCII11b]
    INOUE, Masashi; MATSUDA, Takuya and YOKOYAMA, Shoichi
    "Web Resource Selection for Dialogue System Generating Natural Responses"(poster)
    HCI International 2011 -- Posters' Extended Abstracts, Orlando, FL, 12-14 July 2011.
    Communications in Computer and Information Science, volume 173, pp.571-575, Springer
    [@Springer] [author-version(pdf)] [Google Scholar Citation]

    - This paper proposes the use of Web BBS as the source of utterence examples for an example-based chatbot.
    There are several approaches to change written text into colloquial utterences.
    However, we argue that by exploiting the quasi-spoken web text, chatbots become reasonably natural and more entertaining.
    - The APIs used in this paper for keyword scoring and web searching are no longer available.
    [ICIS10]
    INOUE, Masashi and NGUYEN, Manh Hong
    "User-model-based Evaluation for Interactive Image Retrieval"(poster)
    9th IEEE/ACIS International Conference on Computer and Information Science (ICIS 2010), pp.538-539, Kaminoyama Japan, Aug. 2010.
    [Xplore] [IEEE CSDL] [author-version(pdf)]

    - This paper proposes a framework to automatically evaluate interactive image retrieval systems by simulation using log-based user models.
    [MUST10]
    INOUE, Masashi; OGIRHARA, Mitsunori; HANADA, Ryoko; FURUYAMA, Nobuhiro
    "Utility of Gestural Cues in Indexing Semantic Miscommunication"
    The 2010 International Workshop on Multimedia and Semantic Technologies (MUST 2010), pp.1-6, Busan Korea, May 2010.
    [author-version(pdf)] [Xplore] [Google Scholar Citation]

    -Please see [MTAP11] which is more detailed on this topic.
    [LNCS09CLEF08]
    INOUE, Masashi and GROVER, Piyush
    "Query Types and Visual Concept-based Post-retrieval Clustering"
    Evaluating Systems for Multilingual and Multimodal Information Access
    Vol. LNCS 5706, pp. 661-668, Springer, Sept. 2009. [@Springer] [pdf (author-version)] [Google Scholar Citation]
    --- revised digest of [CLEF08WN]

    - This paper discussed the query topic dependency of the utility of output POST-clustering algorithms in image retrieval for result diversity and re-ranking.
    [HCC08]
    SCHERP, Ansgar; NACK, Frank; NAHRSTEDT, Klara; INOUE, Masashi, GIRGENSOHN, Andreas; HENRICH, Andreas; SANDHAUS, Philipp; THIEME, Sabine; ZHOU, Michelle
    "Interaction and User Experiences with Multimedia Technologies--Challenges and Future Topics"
    3rd International Workshop on Human-Centered Computing -- in conjunction with ACM Multimedia 2008, pp. 1-6, Vancouver, Canada, October 31, 2008.
    [@ACM] [Google Scholar Citation]

    - In section 2, I argued that interactivity is costly and interaction in information access should be minimized.
    [LNCS07CLEF06]
    INOUE, Masashi
    "Using Visual Linkages for Multilingual Image Retrieval"
    Evaluation of Multilingual and Multi-modal Information Retrieval: 7th Workshop of the Cross-Language Evaluation Forum
    Vol. LNCS 4730, pp. 617-624, Springer, Sept. 2007. [@Springer] [pdf (author-version)]
    --- revised digest of [CLEF06WN]

    - This paper report the results of the late-fusion method of textual and visual features on the ImageCLEFphoto ad-hoc task by means of PRE-processing.
    [DMIR07]
    INOUE, Masashi
    "Mining Visual Knowledge for Multi-Lingual Image Retrieval"
    IEEE International Symposium on Data Mining and Information Retrieval (DMIR-07)
    in conjunction with The IEEE 21st International Conference on Advanced Information Networking and Applications (AINA-07) , pp. 307-312, May 21-23, 2007, Sheraton Fallsview Hotel, Niagara Fall, Ontario, Canada
    [@IEEE-Xplore(PDF,BibTex)] [@IEEE-CS(PDF,BibTex)] [pdf (author-version)] [Google Scholar Citation]

    - This paper proposed the use of micro-clustering to enhance visual result diversity (diversification) and re-ranking in image retrieval.
    [LNCS06CLEF05]
    INOUE, Masashi
    "Easing Erroneous Translations in Cross-Language Image Retrieval using Word Associations"
    Accessing Multilingual Information Repositories: 6th Workshop of the Cross-Language Evaluation Forum, Vol. LNCS 4022, pp. 582-591, Springer, 2006.
    [@Springer] [pdf (author-version)] [Google Scholar Citation]
    --- revised digest of [CLEF05WN]
    [MLIA06]
    INOUE, Masashi
    "The Remarkable Search Topic-Finding Task to Share Success Stories of Cross-Language Information Retrieval"
    SIGIR 2006 Workshop: New Directions In Multilingual Information Access, August 11, Seattle, 2006, pp. 61-64.
    [pdf (author-version)] [Google Scholar Citation]

    - When is CLIR useful in our information lives? Can computers answer this question?
    In this paper, I proposed a automated search scenario finding task in which multilingual information access is useful.
    [SAC-IAR05]copyright
    INOUE, Masashi and UEDA, Naonori
    "Retrieving Lightly Annotated Images using Image Similarities"
    the ACM Symposium on Applied Computing (SAC), Special Track on Information Access and Retrieval (IAR), Santa Fe, March 14-17, 2005, pp. 1031-1037
    [@ACM(PDF,BibTex)] [pdf (author-version)] [Google Scholar Citation]

    - Most images are often comes with no or little textual information. This paper proposed a method to build a retrieval model when textual annotations are scarese. The method does not use the annotation data as lightly supervised or lightly labeled data for image indexing, but retrieval models are estimated in unsupervised manner by combining textual and visual information.
    [IRiX04]
    INOUE, Masashi
    "On the need for annotation-based image retrieval"
    Information Retrieval in Context (IRiX): A Workshop at SIGIR 2004(IRiX) Sheffield, UK, 29th July 2004. pp. 44-46
    [pdf (author-version)] [Google Scholar Citation]
    The entire workshop working notes (pdf) and my slide (pdf)

    - This presentation argues that the annotations associated with images are representations of "context in creation" rather than "context in use". Therefore, annotation-based image retrieval (ABIR) or text-based image retrieval can perform context-based retrieval that content-based image retrieval (CBIR) cannot perform. Additionally, the presentation suggests two major research tasks in ABIR.
    - The statement "Thus, QbT is a mono-medium retrieval." in the first page should be "Thus, QbE is a mono-medium retrieval." (Thanks to Rieks Op den Akker).
    [NNSP2001]copyright
    INOUE, Masashi and UEDA, Naonori
    "HMMs for both Labeled and Unlabeled Time Series Data"
    11th IEEE workshops on Neural Networks for Signal Processing,
    Falmouth, Massachusetts, USA, Sept. 10-12, 2001. (NNSP2001)
    [abstract (html)] [@IEEE-Xplore(PDF,BibTex)] [pdf (author-version)] [Google Scholar Citation]
    --- This presentation was supported by "Inoue Foundation for Science".
    NNSP is renamed to IEEE International Workshop on Machine Learning for Signal Processing(MLSP)


    - Please see [PAMI03] for updated results on this topic.
    This article appears in:
    Neural Networks for Signal Processing XI, 2001.
    - Proceedings of the 2001 IEEE Signal Processing Society Workshop, pp. 93-102.

Presentations without proceedings:

    [JADH16]
    INOUE, Masashi
    "Relationship between film information and audience measurement at a film festival"
    [abstract(PDF)]
    Japanese Association for Digital Humanities Conference 2016
    Tokyo, Japan. Sep 12-14, 2016
    Summary: This work reports statistical relationships between properties of screened films and audience behavior in Yamagata International Documentary Film Festival (YIDFF).
    [ISGS10]
    INOUE, Masashi; FURUYAMA, Nobuhiro; Hanada, Ryoko; IRINO, Toshio; MASSAKI, Hiroyasu; and ICHINOMIYA, Takako
    "Successful head-nodding movements in psychotherapeutic process -- when and how"
    p. 193 (abstract only)[pdf]
    4th Conference of the International Society for Gesture Studies (ISGS)
    Frankfurt/Oder, Germany, July 25-30, 2010
    [CSFP08a]
    INOUE, Masashi; HANADA, Ryoko; and FURUYAMA, Nobuhiro
    "Assessment of counseling stages to better understand the process toward solutions"
    2008 Conference on Solution Focused Practice, Austin, TX, November 14 - 15, 2008
    Summary: Although therapists should focus on solutions during counseling sessions, there is a need to talk about other issues as preparation or exploration to bring enough contexts toward the solution. In this study, we segment counseling sessions into several stages and examine their roles and constituent factors.
    [CSFP08b]
    HANADA, Ryoko; INOUE, Masashi; and FURUYAMA, Nobuhiro
    "Successful and unsuccessful 'compliment' in Japanese: The cultural perspective of the communication in solution focused brief therapy"
    2008 Conference on Solution Focused Practice, Austin, TX, November 14 - 15, 2008
    Summary: This article discusses the characteristics of the communication between the therapist and the client on the successful and unsuccessful "compliment", the technique used in solution focused brief therapy in Japanese. We argue from a cultural viewpoint about difficulties of "compliment" due to humbleness of Japanese.
    [CSFP08c]
    FURUYAMA, Nobuhiro; HANADA, Ryoko; and INOUE, Masashi
    Conflict between what the Client says and what he conveys with gesture. ---A Study on the dialogue about "Scaling" in Intervention of solution focused brief therapy
    2008 Conference on Solution Focused Practice, Austin, TX, November 14 - 15, 2008
    Summary: We report on a case study, examining the dialogue between the therapist and the client on the meaning of the minimum and the maximum values of "scaling," the technique used in solution focused brief therapy, with reference to manual gestures that spatially represent the scale.
    [ISGS07]
    HANADA, Ryoko; FURUYAMA, Nobuhiro; and INOUE, Masashi
    "Speech-gesture mismatch and how it changes in the problem description by a client in psychotherapy"
    pp. 5-6 (abstract only)[pdf]
    International Society for Gesture Studies: Third International Conference
    Evanston, Illinois, June 18-21, 2007
    [HBM03]
    INOUE, Masashi; HIGUCHI, Satomi; KAWAWAKI, Dai; KANEKO, Yuko; and NIKI, Chiharu
    "Reading of Handwritten and Printed Text: An fMRI Study"
    The Ninth Annual Meeting of the Organization for Human Brain Mapping (HBM2003)
    New York, USA, June 19-22 2003. [abstract(#1296)] [poster(pdf)]
    --- This presentation was supported by "Foundation for C&C Promotion"
    This presentation discusses the role of left fusiform gyrus and the left lingual gyrus that activated consistently as the visual word form area (VWFA). Activations in right lingual gyrus is varied and associated with learning effects.
    The official abstract is available on CD-Rom in NeuroImage Vol. 19, No. 2, 2003
    as Presented at the 9th International Conference on Functional Mapping of the Human Brain

Other documents:

    [NTCIR16STC]
    Hiroshi UENO, Takya YABUKI, Masashi INOUE
    "YUILA at the NTCIR-12 Short Text Challenge: Combining Twitter Data with Dialogue System Logs"
    Working Note [pdf]
    The 12th NTCIR Short Text Conversation (STC) Japanese subtask
    June 7-10, 2016, Tokyo, Japan
    - This working note describes our approach to solving the responsiveness problem in STC task by using external dialogue log corpus between human and a dialogue system instead of Twitter post-reply pairs.
    [Dagstuhl08]
    INOUE, Masashi
    "Difficulties in image retrieval"
    Dagstuhl Seminar: Contextual and Social Media Understanding and Usage 15-20 June, 20008
    Dagstuhl Seminar Proceedings 08251
    [abstract, pdf, bibtex@Drops]

    The semantic gap is often regarded as a major problem in the field of image retrieval research. In this paper, I will show that there are other important topics that should be addressed for improving the image retrieval utility. Among them, the exploitation of limited information and motivating the use of images are considered to be central to the development of image retrieval.
    [CLEF08WN]
    INOUE, Masashi; GROVER, Piyush
    "Effects of Visual Concept-based Post-retrieval Clustering in ImageCLEFphoto 2008" [pdf] [Google Scholar Citation]
    (Working Note for the ImageCLEFphoto task)
    CLEF2008 17-19 September, Aarhus, Denmark, 2008
    [CLEF06WN]
    INOUE, Masashi
    "Visual Micro-clustering Pre-processing for Cross-Language Ad Hoc Image Retrieval" [pdf]
    (Working Note for the ImageCLEFphoto task)
    - Please see [LNCS07CLEF06] for the conference paper versoin.
    CLEF2006 20 - 22 September, Alicante, Spain, 2006
    [CLEF05WN]
    INOUE, Masashi
    "Recovering translation errors in cross language image retrieval by word association models" [pdf]
    (Working Note for the ImageCLEF ad hoc task)
    CLEF2005 21-23 September, Vienna, Austria, 2005
    [PhD Thesis]
    INOUE, Masashi
    "Statistical learning from Multiple Information Sources"
    Doctor's Thesis, NAIST-DT0161006, March 2004. [web version (pdf)]
    - Appendix 1 explains the difference between Mixture of Experts (MOE) models and Mixture of Generative models [appendix 1 only (pdf)]






ACM Copyright Notice
"(c) ACM. This is the author's version of the work. It is posted here by permission of ACM for your personal use. Not for redistribution. The definitive version may be obtained from ACM Portal.
IEEE Copyright Notice
This material is presented to ensure timely dissemination of scholarly and technical work. Copyright and all rights therein are retained by authors or by other copyright holders. All persons copying this information are expected to adhere to the terms and constraints invoked by each author's copyright. In most cases, these works may not be reposted without the explicit permission of the copyright holder.
The definitive version of the article may be obtained from Xplore.