¥¼¥ß¥Ê¡¼¥ëȯɽ

Æü»þ: 09·î28Æü¡ÊÌÚ¡Ë4¸Â¡Ê15:10-16:40¡Ë


²ñ¾ì: L1

»Ê²ñ: ¿Ûˬ Çîɧ
¿ù»³¡¡µý»Öϯ 1661011: D, Ãæ´Öȯɽ ÃÎǽ¥³¥ß¥å¥Ë¥±¡¼¥·¥ç¥ó Ãæ¼ ů, ¾¾ËÜ Íµ¼£, µÈÌî ¹¬°ìϺ, ÅÄÃæ ¹¨µ¨, ¿ÜÆ£ ¹î¿Î
title: The dialog system regarding consistency
abstract: Contradictory utterances from dialog systems may cause dialog breakdown. If systems have consistency, users will talk to them repeatedly with good feelings. In this study, firstly, I collected contradictory utterances yielded by a dialog system. Then, I investigated what kind of contradictions are occurred by the system. Finally, I considered how to prevent generating contradictory utterances.
language of the presentation: Japanese
ȯɽÂêÌÜ: ÂÐÏäÎÀ°¹çÀ­¤ò¹Íθ¤·¤¿ÂÐÏÃ¥·¥¹¥Æ¥à
ȯɽ³µÍ×: ÂÐÏäÎÃæ¤ÇÌ·½â¤¹¤ëȯÏäòÂÐÏÃ¥·¥¹¥Æ¥à¤¬À¸¤¸¤ë¤³¤È¤Ï¡¢¤½¤ÎÂÐÏäòÇËþ¤µ¤»¡¢ÂÐÏäηѳ¤äÂÐÏÃÁê¼ê¤È¤Î´Ø·¸¹½ÃÛ¤ò˸¤²¤ëÍ×°ø¤È¤Ê¤ë¡£ ¥æ¡¼¥¶¤Ë¹¥´¶¤ò»ý¤Ã¤Æ»È¤Ã¤Æ¤â¤é¤¦¤¿¤á¤Ë¤Ï¡¢Ì·½â¤·¤¿È¯Ïäò¹Ô¤ï¤Ê¤¤¡¢À°¹çÀ­¤Î¤¢¤ëÂÐÏÃ¥·¥¹¥Æ¥à¤Ç¤¢¤ë¤³¤È¤¬µá¤á¤é¤ì¤ë¡£ Ëܸ¦µæ¤Ç¤Ï¡¢¼ÂºÝ¤Ë»¨ÃÌÂÐÏÃ¥·¥¹¥Æ¥à¤¬À¸¤¸¤¿Ì·½â¤¹¤ëȯÏäò¼ý½¸¤·¡¢¤É¤Î¤è¤¦¤ÊÌ·½â¤¬µ¯¤³¤Ã¤Æ¤¤¤ë¤«¤òÄ´ºº¤·¤¿¡£ ¤Þ¤¿¡¢Ì·½â¤¹¤ëȯÏäòËɤ°¤¿¤á¤Î¼êË¡¤ò¸¡Æ¤¤¹¤ë¡£
 
Ë­Åè¡¡¾Ï¹¨ 1651078: M, 2²óÌÜȯɽ ÃÎǽ¥³¥ß¥å¥Ë¥±¡¼¥·¥ç¥ó Ãæ¼ ů, ¾¾ËÜ Íµ¼£, µÈÌî ¹¬°ìϺ, ¿ÜÆ£ ¹î¿Î, ÅÄÃæ ¹¨µ¨
title: Automatic detection of topic change breakdown of non-task-oriented dialogue system
abstract: The non-task-oriented dialogue system has a problem that the system is often not able to change a topic naturally. In other words, the system sometimes generate utterances that conflicts with the dialogue history. This study focuses on automatic detection of topic change breakdown of non-task-oriented dialogue system. In particular, the system evaluates the naturalness of topic transition of the system utterance when the system decides to change a topic. The evaluation method make it possible to reduce a number of topic change breakdown because the system selects an appropriate utterance by using the evaluation of naturalness to the dialogue history. Especially, we focus on the evaluation of topic transition using change between vectors of utterances in latent spaces.
language of the presentation: Japanese
ȯɽÂêÌÜ: »¨ÃÌÂÐÏÃ¥·¥¹¥Æ¥à¤ÎÇËþ¤·¤¿ÏÃÂêÁ«°Ü¤Î¼«Æ°¸¡½Ð
ȯɽ³µÍ×: »¨ÃÌÂÐÏÃ¥·¥¹¥Æ¥à¤Ç¤Ï¡¢1¤Ä¤ÎÂÐÏäǰ·¤¦ÂоݤȤʤë¥É¥á¥¤¥ó¤¬Â¿´ô¤ËÅϤꡢ¤·¤Ð¤·¤ÐÂÐÏÃÃæ¤Î¥É¥á¥¤¥óÁ«°Ü¤Ë¤ª¤¤¤Æ¡¢¥·¥¹¥Æ¥à¤¬À¸À®¤·¤¿È¯Ïä¬ÂÐÏÃÍúÎò¤ÎÏÃÂê¤È·Ò¤¬¤é¤º¼«Á³¤ÊÏÃÂêÁ«°Ü¤¬¤Ç¤­¤Ê¤¤ÌäÂ꤬µ¯¤³¤ë¡£ ¤½¤³¤ÇËܸ¦µæ¤Ç¤Ï¡¢»¨ÃÌÂÐÏÃ¥·¥¹¥Æ¥à¤Ë¤ª¤±¤ëÇËþ¤·¤¿ÏÃÂêÁ«°Ü¤Î¼«Æ°¸¡½Ð¤Ë¼è¤êÁȤó¤À¡£¶ñÂÎŪ¤Ë¤Ï¡¢¤¢¤ë¥æ¡¼¥¶È¯ÏäËÂФ·¤Æ¥·¥¹¥Æ¥àȯÏä¬À¸À®¤µ¤ì¤¿ºÝ¤Ë¡¢¤É¤ÎÄøÅÙ¼«Á³¤ËÏÃÂêÁ«°Ü¤¬¤Ç¤­¤Æ¤¤¤ë¤«¤òɾ²Á¤¹¤ë¡£ÂÐÏÃÁ«°Ü¤Î¼«Á³¤µ¤òɾ²Á¤¹¤ë¥â¥Ç¥ë¤ò¹½ÃÛ¤¹¤ë¤³¤È¤Ç¡¢ÂÐÏÃÍúÎò¤ËÂФ·¤ÆÂÅÅö¤Ç¤¢¤ëÂÐÏÃ¥·¥¹¥Æ¥à¤ÎȯÏøõÊä¤òÁªÂò¤¹¤ë¤³¤È¤¬²Äǽ¤È¤Ê¤ê¡¢ÇËþ¤·¤¿ÏÃÂêÁ«°Ü¤ÎÍÞÀ©¤¬¤Ç¤­¤ë¤È¹Í¤¨¤é¤ì¤ë¡£¤³¤Î¤¿¤áÆäˡ¢È¯ÏÃ¥Ù¥¯¥È¥ë¤ÎÀøºß¶õ´Ö¤Ë¤ª¤±¤ëÊѲ½¤òÍѤ¤¤¿ÏÃÂêÁ«°Ü¤Îɾ²Á¤Ë¤Ä¤¤¤Æ¸¡Æ¤¤·¤Æ¤¤¤ë¡£
 
¾¾ÅÄ¡¡µÁµ® 1651099: M, 2²óÌÜȯɽ ÃÎǽ¥³¥ß¥å¥Ë¥±¡¼¥·¥ç¥ó Ãæ¼ ů, °ÂËÜ ·Ä°ì, ÎëÌÚ Í¥, ÅÄÃæ ¹¨µ¨
title: Trade-off between task complexity and estimating worker's ability by worker's behavior in croudsourcing
abstract: In crowdsourcing, increasing reliable workers leads to high quality products and cost reduction.However, malicious workers or workers who have no ability to work accurately sometimes work.It is important to estimate the worker's ability in order to eliminate them or provide guidance for them.There are some studies to estimate the worker's ability by the worker's behavior.It is difficult to apply these studies to simple tasks because behavior that can be acquired is limited.In this research, we aim to improve the accuracy of estimating workers' abilities by complicating simple tasks and acquiring many behaviors.Furthermore, we are planning to verify the trade-off between the acquiring effective behavior for estimating the worker 's ability and the burden on the workers.
language of the presentation: Japanese
 
¾®ËÒ¡¡¿¿»Ò 1651048: M, 2²óÌÜȯɽ ÃÎǽ¥³¥ß¥å¥Ë¥±¡¼¥·¥ç¥ó Ãæ¼ ů, °ÂËÜ ·Ä°ì, ÎëÌÚ Í¥, ÅÄÃæ ¹¨µ¨
title:Work scheduling subdivision method using agricultural data
abstract:In this research, we aim to refine the work schedule in agriculture.We propose a model that can aggregate time series data of agricultural work for each field and can standardize work procedure and evaluation method. There are uncertainties such as weather and environment for each field when extracting work procedures from farm work data.If using mathematical programming method for schedule refinement by this uncertainty factor, There is a problem that the calculation amount becomes enormous and it becomes difficult to select a variable. In this method, by extracting common data of tasks performed in a plurality of fields, the work procedure is formulated taking into consideration unspecified factors. By doing this, detailed work procedures can be extracted even in the presence of uncertain factors that were problems. In this presentation we will describe the above contents and experimental results as a foothold.
language of the presentation: Japanese
ȯɽÂêÌÜ:ÇÀºî¶È¥Ç¡¼¥¿¤òÍѤ¤¤¿ºî¶È¥¹¥±¥¸¥å¡¼¥ëºÙʬ²½¼êË¡
ȯɽ³µÍ×:Ëܸ¦µæ¤Ç¤Ï¡¤ÇÀ¶È¤Ë¤ª¤±¤ëºî¶È¥¹¥±¥¸¥å¡¼¥ë¤Î¾ÜºÙ²½¤òÌÜŪ¤È¤¹¤ë. ÇÀºî¶È¤Î»þ·ÏÎó¥Ç¡¼¥¿¤òȪ¤´¤È¤Ë½¸Ìó¤·¡¢ºî¶È¼ê½ç¤òÄê·¿²½¤Ç¤­¤ë¥â¥Ç¥ë¤Èɾ²ÁÊýË¡¤òÄó°Æ¤¹¤ë. ÇÀºî¶È¤Î¥Ç¡¼¥¿¤«¤éºî¶È¼ê½ç¤òÃê½Ð¤¹¤ëºÝÅ·¸õ¤äȪ¤´¤È¤Î´Ä¶­¤È¤¤¤Ã¤¿ÉÔ³ÎÄêÍ×ÁǤ¬Â¸ºß¤¹¤ë.¤³¤ÎÉÔ³ÎÄêÍ×ÁǤˤè¤Ã¤Æ¥¹¥±¥¸¥å¡¼¥ë¾ÜºÙ²½¤Î¤¿¤á¤Î¿ôÍý·×²èË¡¤òÍѤ¤¤¿¾ì¹ç, ·×»»Î̤¬ËÄÂç¤È¤Ê¤ë¤³¤È¤äÊÑ¿ôÁªÂò¤¬Æñ¤·¤¯¤Ê¤ë¤È¤¤¤¦ÌäÂ꤬¤¢¤ë¡£ ËܼêË¡¤Ç¤ÏÊ£¿ô¤ÎȪ¤Ç¹Ô¤ï¤ì¤¿ºî¶È¤Î¶¦Ä̤¹¤ë¥Ç¡¼¥¿¤òÃê½Ð¤¹¤ë¤³¤È¤Ç,ÉÔÆÃÄêÍ×ÁǤò¹Íθ¤·¤¿¾å¤Çºî¶È¼ê½ç¤òÄê·¿¤¹¤ë. ¤³¤ì¤Ë¤è¤ê,ÌäÂê¤Ç¤¢¤Ã¤¿ÉÔ³ÎÄêÍ×ÁǤ¬Â¸ºß¤¹¤ëÃæ¤Ç¤¢¤Ã¤Æ¤â¾ÜºÙ¤Êºî¶È¼ê½ç¤ÎÃê½Ð¤¬²Äǽ¤È¤Ê¤ë. ËÜȯɽ¤Ç¤Ï¾åµ­¤ÎÆâÍƤÈ­¤¬¤«¤ê¤È¤Ê¤ë¼Â¸³·ë²Ì¤Ë¤Ä¤¤¤Æ½Ò¤Ù¤ë.
 

²ñ¾ì: L2

»Ê²ñ: Gao Juntao
´ä°æ¡¡ÍºÂç 1651018: M, 2²óÌÜȯɽ ÃÎǽ¥·¥¹¥Æ¥àÀ©¸æ ¿ùËÜ ¸¬Æó, °ÂËÜ ·Ä°ì, ¾¾¸¶ ¿ò½¼, Æî ͵¼ù(µÒ°÷¡Ë, ¾®Â¢ Àµµ±, ¾®ÎÓ ÂÙ²ð
title: Prediction Governor for Input Affine Nonlinear Systems and Its Application to Automatic Cruise Control
abstract: In the automatic driving system, prediction accuracy of the driving route is very important to achieve high performance automatic driving. As a method of improving prediction accuracy, the previous work presented the prediction governor which shapes predicted data. However, the previous result cannot applied to nonlinear systems as automobiles. Motivated by this, this study addresses the design problem of prediction governors for input affine nonlinear systems. Then, we analytically derive an optimal prediction governor which is characterized by plant models. Furthermore, we apply it to the automatic cruise control problem, and its usefulness is verified by experiment.
language of the presentation: Japanese
ȯɽÂêÌÜ: ÆþÎÏ¥¢¥Õ¥£¥óÈóÀþ·Á¥·¥¹¥Æ¥à¤ËÂФ¹¤ëͽ¬¥¬¥Ð¥Ê¤ÎÀ߷פȼ«Æ°Áö¹ÔÀ©¸æ¤Ø¤Î±þÍÑ
ȯɽ³µÍ×: ¼«Æ°±¿Å¾¥·¥¹¥Æ¥à¤Ë¤ª¤¤¤Æ¡¤¹âÀ­Ç½¤Ê¼«Æ°Áö¹Ô¤ò¼Â¸½¤¹¤ë¤¿¤á¤Ë¤ÏÁö¹Ô¥ë¡¼¥È¤Îͽ¬ÀºÅÙ¤¬½ÅÍפȤʤ롥ͽ¬ÀºÅÙ¤ò¸þ¾å¤µ¤»¤ëÊýË¡¤È¤·¤Æ¡¤Àè¹Ô¸¦µæ¤Ç¤Ïͽ¬ÃͤòÀ°·Á¤¹¤ëͽ¬¥¬¥Ð¥Ê¤¬Äó°Æ¤µ¤ì¤Æ¤¤¤ë¡¥¤·¤«¤·¡¤Í½Â¬¥¬¥Ð¥Ê¤Ï¼«Æ°¼Ö¤Î¤è¤¦¤ÊÈóÀþ·Á¥·¥¹¥Æ¥à¤ËŬÍѤ¹¤ë¤³¤È¤¬¤Ç¤­¤Ê¤¤¡¥¤½¤Î¤¿¤á¡¤Ëܸ¦µæ¤Ç¤ÏÈóÀþ·Á¥·¥¹¥Æ¥à¤ËŬÍѲÄǽ¤Êͽ¬¥¬¥Ð¥Ê¤òÀ߷פ¹¤ë¡¥¤µ¤é¤Ë¡¤Äó°Æ¤¹¤ëͽ¬¥¬¥Ð¥Ê¤ò¼«Æ°±¿Å¾¥·¥¹¥Æ¥à¤ËŬÍѤ·¡¤¼Âµ¡¼Â¸³¤òÄ̤·¤Æ¤½¤ÎÍ­ÍÑÀ­¤ò¸¡¾Ú¤¹¤ë¡¥
 
üâËÜ¡¡ÍºÂÀ 1651068: M, 2²óÌÜȯɽ ÃÎǽ¥·¥¹¥Æ¥àÀ©¸æ ¿ùËÜ ¸¬Æó, °ÂËÜ ·Ä°ì, ¾¾¸¶ ¿ò½¼, Æî ͵¼ù(µÒ°÷¡Ë, ¾®Â¢ Àµµ±, ¾®ÎÓ ÂÙ²ð
title: Development of autonomous distributed lighting control system
abstract: Improvement of lighting environment is an important subject in promoting energy saving. In this research, we propose a programmable lighting environment control system composed of multiple lights and windows that can partially change the transparency. Proposed system is able to control the distribution of illuminance by turning ON/OFF and switching to shaded/transparent modes. At this time, in order to realize the desired illuminance distribution, it is necessary to obtain a controller which determines a proper Lighting/Shading pattern. In this research, we formulated the Lighting/Shading pattern control problem of the proposed system and constructed an autonomous distributed control algorithm. In addition, we verified its effectiveness by performing an experiment.
language of the presentation: Japanese
ȯɽÂêÌÜ: ¼«Î§Ê¬»¶·¿¸÷´Ä¶­À©¸æ¥·¥¹¥Æ¥à¤Î¹½ÃÛ
ȯɽ³µÍ×: ²÷Ŭ¤Ê¸÷´Ä¶­¤Î¼Â¸½¤ä¾Ê¥¨¥Í¥ë¥®¡¼²½¤Î¿ä¿Ê¤Ë¤ª¤¤¤Æ¡¤¾ÈÌÀ´Ä¶­¤Î²þÁ±¤Ï½ÅÍפʲÝÂê¤Ç¤¢¤ë¡¥Ëܸ¦µæ¤Ç¤Ï¡¤Ê£¿ô¤Î¾ÈÌÀ¤È¡¤ÉôʬŪ¤Ë¼×¸÷¤¬²Äǽ¤ÊÁë(¥·¥§¡¼¥É)¤«¤é¹½À®¤µ¤ì¤ë¡¤¥×¥í¥°¥é¥Þ¥Ö¥ë¸÷´Ä¶­À©¸æ¥·¥¹¥Æ¥à¤òÄó°Æ¤¹¤ë¡¥Äó°Æ¥·¥¹¥Æ¥à¤Ï¡¤¾ÈÌÀ¤ÎÅÀÅô/¾ÃÅô¤È¡¤¥·¥§¡¼¥É¤Î¼×¸÷/Æ©¸÷¥Ñ¥¿¡¼¥ó¤òÀÚ¤êÂؤ¨¤ë¤³¤È¤Ë¤è¤Ã¤Æ¡¤¾ÈÅÙʬÉÛ¤òÀ©¸æ¤¹¤ë¤³¤È¤¬¤Ç¤­¤ë¡¥Äó°Æ¥·¥¹¥Æ¥à¤Ë¤ª¤¤¤Æ½ê˾¤Î¾ÈÅÙʬÉÛ¤ò¼Â¸½¤¹¤ë¤¿¤á¤Ë¤Ï¡¤Å¬ÀÚ¤ÊÅÀÅô/¼×¸÷¥Ñ¥¿¡¼¥ó¤ò·èÄꤹ¤ëÀ©¸æ´ï¤òµá¤á¤ëɬÍפ¬¤¢¤ë¡¥Ëܸ¦µæ¤Ç¤Ï¡¤ÅÀÅô/¼×¸÷¥Ñ¥¿¡¼¥óÀ©¸æÌäÂê¤ÎÄê¼°²½¤È¡¤Å¬ÀÚ¤ÊÅÀÅô/¼×¸÷¥Ñ¥¿¡¼¥ó¤ò·èÄꤹ¤ëʬ»¶À©¸æ¥¢¥ë¥´¥ê¥º¥à¤Î¹½ÃÛ¤ò¹Ô¤¦¡¥¤Þ¤¿¡¤¾ÈÅÙ¾ìÀ©¸æ¼Â¸³¤ò¹Ô¤¤¡¤Äó°Æ¼êË¡¤ÎÍ­¸úÀ­¤ò³Îǧ¤¹¤ë¡¥
 
°æ¾å¡¡·ò¸ã 1651014: M, 2²óÌÜȯɽ ÃÎǽ¥·¥¹¥Æ¥àÀ©¸æ ¿ùËÜ ¸¬Æó, ²¬ÅÄ ¼Â, ¾¾¸¶ ¿ò½¼, Æî ͵¼ù(µÒ°÷¡Ë, ¾®Â¢ Àµµ±, ¾®ÎÓ ÂÙ²ð
title:Target value shaping technology in servo system and its application to high frequency power supply system
abstract:In the servo system, it is important to make the output follow the input target value.In this research, we propose a method to shape the target value by using the past information in order to follow the target value.In addition, we consider application of the proposed method to high frequency power source, and verify its usefulness by simulation.
ȯɽÂêÌÜ¡§¥µ¡¼¥Ü·Ï¤Ë¤ª¤±¤ëÌÜɸÃÍÀ°·Áµ»½Ñ¤È¤½¤Î¹â¼þÇÈÅŸ»¥·¥¹¥Æ¥à¤Ø¤Î±þÍÑ
³µÍס§¥µ¡¼¥Ü·Ï¤Ç¤Ï¡¢ÆþÎϤ·¤¿ÌÜɸÃͤ˽ÐÎϤòÄɽ¾¤µ¤»¤ë¤³¤È¤¬½ÅÍפǤ¢¤ë¡£Ëܸ¦µæ¤Ç¤Ï¡¢ÌÜɸÃÍÄɽ¾¤ò¹Ô¤¦¤¿¤á¤Ë¡¢²áµî¤Î¾ðÊó¤òÍøÍѤ·¤ÆÌÜɸÃͤòÀ°·Á¤¹¤ëÊýË¡¤òÄó°Æ¤¹¤ë¡£ ¡¡¡¡¡¡¤Þ¤¿¡¢Äó°Æ¼êË¡¤Î¹â¼þÇÈÅŸ»¤Ø¤Î±þÍѤò¹Í¤¨¡¢¥·¥ß¥å¥ì¡¼¥·¥ç¥ó¤Ë¤è¤ê¤½¤ÎÍ­ÍÑÀ­¤ò¸¡¾Ú¤¹¤ë
 
RODRIGUEZ RAMIREZ JUAN ESTEBAN 1661031: D, Ãæ´Öȯɽ ÃÎǽ¥·¥¹¥Æ¥àÀ©¸æ ¿ùËÜ ¸¬Æó, ³Þ¸¶ Àµ¼£, ¾¾¸¶ ¿ò½¼, Æî ͵¼ù(µÒ°÷¡Ë, ¾®Â¢ Àµµ±, ¾®ÎÓ ÂÙ²ð
title: Design of practical dynamic quantizers for networked control systems
abstract: Quantization is a very important process in the operation of networked control systems (NCSs). Apart from the traditional quantization related problems, like the performance degradation of the system and the data rate constraints of the channel, there are numerous other problems that affect the NCSs and can be solved by the appropriate design of dynamic quantizers. Some of these problems are the reduction of the traffic generated by the system and the design of optimal quantizers when the model of the plant is absent. In this study, two types of quantizers are proposed to tackle a set of specific problems. One is an event-triggered quantizer, and the other is a neural network quantizer. With numerical examples the effectiveness of the proposed quantizers and their design methods are verified.
language of the presentation: English