Shuai Zou, Kento Kuzushima, Hironori Mitake, S. Hasegawa
Recent developments in robotics and virtual reality (VR) are making embodied agents familiar, and social behaviors of embodied conversational agents are essential to create mindful daily lives with conversational agents. Especially, natural nonverbal behaviors are required, such as gaze and gesture movement. We propose a novel method to create an agent with human-like gaze as a listener in multi-party conversation, using Hidden Markov Model (HMM) to learn the behavior from real conversation examples. The model can generate gaze reaction according to users' gaze and utterance. We implemented an agent with proposed method, and created VR environment to interact with the agent. The proposed agent reproduced several features of gaze behavior in example conversations. Impression survey result showed that there is at least a group who felt the proposed agent is similar to human and better than conventional methods.
{"title":"Conversational Agent Learning Natural Gaze and Motion of Multi-Party Conversation from Example","authors":"Shuai Zou, Kento Kuzushima, Hironori Mitake, S. Hasegawa","doi":"10.1145/3125739.3132607","DOIUrl":"https://doi.org/10.1145/3125739.3132607","url":null,"abstract":"Recent developments in robotics and virtual reality (VR) are making embodied agents familiar, and social behaviors of embodied conversational agents are essential to create mindful daily lives with conversational agents. Especially, natural nonverbal behaviors are required, such as gaze and gesture movement. We propose a novel method to create an agent with human-like gaze as a listener in multi-party conversation, using Hidden Markov Model (HMM) to learn the behavior from real conversation examples. The model can generate gaze reaction according to users' gaze and utterance. We implemented an agent with proposed method, and created VR environment to interact with the agent. The proposed agent reproduced several features of gaze behavior in example conversations. Impression survey result showed that there is at least a group who felt the proposed agent is similar to human and better than conventional methods.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"72 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115546488","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
"89:&#;<=$ Dr. Alexander Waibel is a Professor of Computer Science at Carnegie Mellon University, Pittsburgh and at the Karlsruhe Institute of Technology, Germany. He is the director of the International Center for Advanced Communication Technologies (interACT). The Center works in a network with eight of the world’s top research institutions. Its mission is to develop advanced machine learning algorithms to improve human-human and human-machine communication technologies. Prof. Waibel and his team pioneered many statistical and neural learning algorithms that made such communication breakthroughs possible. Most notably, the “Time-Delay Neural Network” (1987) (now also known as “convolutional” neural network) is at the heart of many of today’s AI technologies. System breakthroughs that followed suit included early multimodal dialog interfaces, the first speech translation system in Europe&USA (1990/1991), the first simultaneous lecture interpretation system (2005), and Jibbigo, the first commercial speech translator on a phone (2009).
89:&#;<=$ Alexander Waibel博士是匹兹堡卡内基梅隆大学和德国卡尔斯鲁厄理工学院的计算机科学教授。他是国际先进通信技术中心(interACT)的主任。该中心与世界上八家顶级研究机构建立了一个网络。它的使命是开发先进的机器学习算法，以改善人与人之间和人机通信技术。Waibel教授和他的团队开创了许多统计和神经学习算法，使这种通信突破成为可能。最值得注意的是，“时滞神经网络”(1987年)(现在也被称为“卷积”神经网络)是当今许多人工智能技术的核心。随后的系统突破包括早期的多模态对话界面，欧洲和美国的第一个语音翻译系统(1990/1991)，第一个同声传译系统(2005)，以及Jibbigo，第一个商业电话语音翻译器(2009)。
{"title":"Keynote Talk","authors":"A. Waibel","doi":"10.1145/3125739.3134523","DOIUrl":"https://doi.org/10.1145/3125739.3134523","url":null,"abstract":"\"89:&#;<=$ Dr. Alexander Waibel is a Professor of Computer Science at Carnegie Mellon University, Pittsburgh and at the Karlsruhe Institute of Technology, Germany. He is the director of the International Center for Advanced Communication Technologies (interACT). The Center works in a network with eight of the world’s top research institutions. Its mission is to develop advanced machine learning algorithms to improve human-human and human-machine communication technologies. Prof. Waibel and his team pioneered many statistical and neural learning algorithms that made such communication breakthroughs possible. Most notably, the “Time-Delay Neural Network” (1987) (now also known as “convolutional” neural network) is at the heart of many of today’s AI technologies. System breakthroughs that followed suit included early multimodal dialog interfaces, the first speech translation system in Europe&USA (1990/1991), the first simultaneous lecture interpretation system (2005), and Jibbigo, the first commercial speech translator on a phone (2009).","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"97 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122580494","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Hao Yin, Keiko Yamamoto, Itaru Kuramoto, Y. Tsujino
With the development of Web 2.0 technology, Q&A websites have become one of the most common avenues for large scale knowledge sharing. However, three types of barriers between questioners and respondents make knowledge sharing difficult:differences in the personality types of the questioners and respondents, lack of trust, and an arrogant or negative attitude exhibited by some questioners. In order to lower these barriers, we propose a Q&A mediator system with virtual character agents. In this system, a questioner asks questions through his/her own agent, and the respondents see the question from their agent. Each agent has four characteristics:personality traits similar to knowledge sharers, good looks, affable personality, and a positive attitude. The results of a preliminary experiment indicated that the proposed system can improve the users' motivation to answer questions.
{"title":"Virtual Character Agent for Lowering Knowledge-sharing Barriers on Q&A Websites","authors":"Hao Yin, Keiko Yamamoto, Itaru Kuramoto, Y. Tsujino","doi":"10.1145/3125739.3132591","DOIUrl":"https://doi.org/10.1145/3125739.3132591","url":null,"abstract":"With the development of Web 2.0 technology, Q&A websites have become one of the most common avenues for large scale knowledge sharing. However, three types of barriers between questioners and respondents make knowledge sharing difficult:differences in the personality types of the questioners and respondents, lack of trust, and an arrogant or negative attitude exhibited by some questioners. In order to lower these barriers, we propose a Q&A mediator system with virtual character agents. In this system, a questioner asks questions through his/her own agent, and the respondents see the question from their agent. Each agent has four characteristics:personality traits similar to knowledge sharers, good looks, affable personality, and a positive attitude. The results of a preliminary experiment indicated that the proposed system can improve the users' motivation to answer questions.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"37 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124932687","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Nikhil Churamani, Paul Anton, M. Brügger, Erik Fließwasser, Thomas Hummel, Julius Mayer, Waleed Mustafa, Hwei Geok Ng, Thi Linh Chi Nguyen, Quan Nguyen, Marcus Soll, S. Springenberg, Sascha S. Griffiths, Stefan Heinrich, Nicolás Navarro-Guerrero, Erik Strahl, Johannes Twiefel, C. Weber, S. Wermter
Advancements in Human-Robot Interaction involve robots being more responsive and adaptive to the human user they are interacting with. For example, robots model a personalised dialogue with humans, adapting the conversation to accommodate the user's preferences in order to allow natural interactions. This study investigates the impact of such personalised interaction capabilities of a human companion robot on its social acceptance, perceived intelligence and likeability in a human-robot interaction scenario. In order to measure this impact, the study makes use of an object learning scenario where the user teaches different objects to the robot using natural language. An interaction module is built on top of the learning scenario which engages the user in a personalised conversation before teaching the robot to recognise different objects. The two systems, i.e. with and without the interaction module, are compared with respect to how different users rate the robot on its intelligence and sociability. Although the system equipped with personalised interaction capabilities is rated lower on social acceptance, it is perceived as more intelligent and likeable by the users.
{"title":"The Impact of Personalisation on Human-Robot Interaction in Learning Scenarios","authors":"Nikhil Churamani, Paul Anton, M. Brügger, Erik Fließwasser, Thomas Hummel, Julius Mayer, Waleed Mustafa, Hwei Geok Ng, Thi Linh Chi Nguyen, Quan Nguyen, Marcus Soll, S. Springenberg, Sascha S. Griffiths, Stefan Heinrich, Nicolás Navarro-Guerrero, Erik Strahl, Johannes Twiefel, C. Weber, S. Wermter","doi":"10.1145/3125739.3125756","DOIUrl":"https://doi.org/10.1145/3125739.3125756","url":null,"abstract":"Advancements in Human-Robot Interaction involve robots being more responsive and adaptive to the human user they are interacting with. For example, robots model a personalised dialogue with humans, adapting the conversation to accommodate the user's preferences in order to allow natural interactions. This study investigates the impact of such personalised interaction capabilities of a human companion robot on its social acceptance, perceived intelligence and likeability in a human-robot interaction scenario. In order to measure this impact, the study makes use of an object learning scenario where the user teaches different objects to the robot using natural language. An interaction module is built on top of the learning scenario which engages the user in a personalised conversation before teaching the robot to recognise different objects. The two systems, i.e. with and without the interaction module, are compared with respect to how different users rate the robot on its intelligence and sociability. Although the system equipped with personalised interaction capabilities is rated lower on social acceptance, it is perceived as more intelligent and likeable by the users.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"23 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-27","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"125575798","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Kenta Takeuchi, Dai Hasegawa, S. Shirakawa, Naoshi Kaneko, H. Sakuta, K. Sumi
In this research, we take a first step in generating motion data for gestures directly from speech features. Such a method can make creating gesture animations for Embodied Conversational Agents much easier. We implemented a model using Bi-Directional LSTM taking phonemic features from speech audio data as input to output time sequence data of rotations of bone joints. We assessed the validity of the predicted gesture motion data by evaluating the final loss value of the network, and evaluating the impressions of the predicted gesture by comparing it with the actual motion data that accompanied the audio data used for input and motion data that accompanied a different audio data. The results showed that the accuracy of the prediction for the LSTM model was better than a simple RNN model. In contrast, the impressions evaluation of the predicted gesture was rated lower than the original and mismatched gestures, although individually some predicted gestures were rated the same degree as the mismatched gestures.
{"title":"Speech-to-Gesture Generation: A Challenge in Deep Learning Approach with Bi-Directional LSTM","authors":"Kenta Takeuchi, Dai Hasegawa, S. Shirakawa, Naoshi Kaneko, H. Sakuta, K. Sumi","doi":"10.1145/3125739.3132594","DOIUrl":"https://doi.org/10.1145/3125739.3132594","url":null,"abstract":"In this research, we take a first step in generating motion data for gestures directly from speech features. Such a method can make creating gesture animations for Embodied Conversational Agents much easier. We implemented a model using Bi-Directional LSTM taking phonemic features from speech audio data as input to output time sequence data of rotations of bone joints. We assessed the validity of the predicted gesture motion data by evaluating the final loss value of the network, and evaluating the impressions of the predicted gesture by comparing it with the actual motion data that accompanied the audio data used for input and motion data that accompanied a different audio data. The results showed that the accuracy of the prediction for the LSTM model was better than a simple RNN model. In contrast, the impressions evaluation of the predicted gesture was rated lower than the original and mismatched gestures, although individually some predicted gestures were rated the same degree as the mismatched gestures.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"91 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115112810","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
In an explorative study concerning the social acceptance of two specific humanoid robots, the experimenter asked participants (N = 36) to place a book in an adjacent room. Upon entering the room, participants were confronted by a NAO or a Pepper robot expressing persistent opposition against the idea of placing the book in the room. On average, 72% of participants facing NAO complied with the robot's requests and returned the book to the experimenter. The corresponding figure for the Pepper robot was 50%, which shows that the two robot morphologies had a different effect on participants' social behavior. Furthermore, results from a post-study questionnaire (GODSPEED) indicated that participants perceived NAO as more likable, intelligent, safe and lifelike than Pepper. Moreover, participants used significantly more positive words and fewer negative words to describe NAO than Pepper in an open-ended interview. There was no statistically significant difference between conditions in participants' negative attitudes toward robots in general, as assessed using the NARS questionnaire.
{"title":"Don't Judge a Book by its Cover: A Study of the Social Acceptance of NAO vs. Pepper","authors":"Sofia Thunberg, Sam Thellman, T. Ziemke","doi":"10.1145/3125739.3132583","DOIUrl":"https://doi.org/10.1145/3125739.3132583","url":null,"abstract":"In an explorative study concerning the social acceptance of two specific humanoid robots, the experimenter asked participants (N = 36) to place a book in an adjacent room. Upon entering the room, participants were confronted by a NAO or a Pepper robot expressing persistent opposition against the idea of placing the book in the room. On average, 72% of participants facing NAO complied with the robot's requests and returned the book to the experimenter. The corresponding figure for the Pepper robot was 50%, which shows that the two robot morphologies had a different effect on participants' social behavior. Furthermore, results from a post-study questionnaire (GODSPEED) indicated that participants perceived NAO as more likable, intelligent, safe and lifelike than Pepper. Moreover, participants used significantly more positive words and fewer negative words to describe NAO than Pepper in an open-ended interview. There was no statistically significant difference between conditions in participants' negative attitudes toward robots in general, as assessed using the NARS questionnaire.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"76 5","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"120838031","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The development of automated technologies in our daily life has transformed the role of human operators from a controller to a teammate who shares control with automated agents. However, this 'teammate' relationship between humans and automation raises an important but challenging research question regarding the formation of human-agent trust. Considering that the formation of human-agent trust is a dynamic and sophisticated process involving human factors, this study conducted a two-phase online experiment to examine personal influences on users' trust propensity and their trust formation in human-agent interactions. Our findings revealed distinctive personal influences on dispositional trust and the formation of human-agent trust at different stages. We found that users who exhibit higher trust propensities in humans also develop higher trust toward automated agents in initial stages. This study, as the first of its kind, not only fills the gap of knowledge about personal influences on human-agent trust, but also offers opportunities to enhance the future design of automated agent systems.
{"title":"Personal Influences on Dynamic Trust Formation in Human-Agent Interaction","authors":"Hsiao-Ying Huang, Masooda N. Bashir","doi":"10.1145/3125739.3125749","DOIUrl":"https://doi.org/10.1145/3125739.3125749","url":null,"abstract":"The development of automated technologies in our daily life has transformed the role of human operators from a controller to a teammate who shares control with automated agents. However, this 'teammate' relationship between humans and automation raises an important but challenging research question regarding the formation of human-agent trust. Considering that the formation of human-agent trust is a dynamic and sophisticated process involving human factors, this study conducted a two-phase online experiment to examine personal influences on users' trust propensity and their trust formation in human-agent interactions. Our findings revealed distinctive personal influences on dispositional trust and the formation of human-agent trust at different stages. We found that users who exhibit higher trust propensities in humans also develop higher trust toward automated agents in initial stages. This study, as the first of its kind, not only fills the gap of knowledge about personal influences on human-agent trust, but also offers opportunities to enhance the future design of automated agent systems.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"64 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"124835886","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
To build a conversational interface wherein an agent system can smoothly communicate with multiple persons, it is imperative to know how the timing of speaking is decided. In this research, we explore the head movements of participants as an easy-to-measure nonverbal behavior to predict the nest-utterance timing, i.e., the interval between the end of the current speaker's utterance and the start of the next speaker's utterance, in turn-changing in multi-party meetings. First, we collected data on participants' six degree-of-freedom head movements and utterances in four-person meetings. The results of the analysis revealed that the amount of head movements of current speaker, next speaker, and listeners have a positive correlation with the utterance interval. Moreover, the degree of synchrony of the head position and posture between the current speaker and next speaker is negatively correlated with the utterance interval. On the basis of these findings, we used their head movements and the synchrony of their head movements as feature values and devised several prediction models. A model using all features performed the best and was able to predict the next-utterance timing well. Therefore, this research revealed that the participants' head movement is useful for predicting the next-utterance timing in turn-changing in multi-party meetings.
{"title":"Prediction of Next-Utterance Timing using Head Movement in Multi-Party Meetings","authors":"Ryo Ishii, Shiro Kumano, K. Otsuka","doi":"10.1145/3125739.3125765","DOIUrl":"https://doi.org/10.1145/3125739.3125765","url":null,"abstract":"To build a conversational interface wherein an agent system can smoothly communicate with multiple persons, it is imperative to know how the timing of speaking is decided. In this research, we explore the head movements of participants as an easy-to-measure nonverbal behavior to predict the nest-utterance timing, i.e., the interval between the end of the current speaker's utterance and the start of the next speaker's utterance, in turn-changing in multi-party meetings. First, we collected data on participants' six degree-of-freedom head movements and utterances in four-person meetings. The results of the analysis revealed that the amount of head movements of current speaker, next speaker, and listeners have a positive correlation with the utterance interval. Moreover, the degree of synchrony of the head position and posture between the current speaker and next speaker is negatively correlated with the utterance interval. On the basis of these findings, we used their head movements and the synchrony of their head movements as feature values and devised several prediction models. A model using all features performed the best and was able to predict the next-utterance timing well. Therefore, this research revealed that the participants' head movement is useful for predicting the next-utterance timing in turn-changing in multi-party meetings.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"44 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"126743558","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
The feeling of being looked back when we look at someone and that someone is also aware that we are looking at him/her is a basic fundamental during social interaction. This situation can only occur if both realize the presence of each other. Based on these theories, this research is motivated in exploiting the possibility of designing for a gaze sensitive object - how people can relate to object by depending on their eyes only. In this paper, we present a gaze-activated coffee machine called the CoffeePet attached with two small, OLED screen that will displays animated eyes. These eyes are responsive towards the user's gaze behavior. Furthermore, we used a sensor module (HVC Omron) to detect and track the eyes of a user in real time. It gives the ability for the user to interact with the CoffeePet simply by moving their eyes. The CoffeePet is also able to automatically brew and pour the coffee out of its spout if it feels appropriate during the interaction. We further explain the description of the system, modification of the real product, and the experimental plan to compare the user's perception of the CoffeePet's eyes and to investigate whether the user realizes or not that their gaze behavior influences the CoffeePet to react.
{"title":"Exploring Gaze-Activated Object With the CoffeePet","authors":"S. A. Anas, G.W.M. Rauterberg, Jun Hu","doi":"10.1145/3125739.3132578","DOIUrl":"https://doi.org/10.1145/3125739.3132578","url":null,"abstract":"The feeling of being looked back when we look at someone and that someone is also aware that we are looking at him/her is a basic fundamental during social interaction. This situation can only occur if both realize the presence of each other. Based on these theories, this research is motivated in exploiting the possibility of designing for a gaze sensitive object - how people can relate to object by depending on their eyes only. In this paper, we present a gaze-activated coffee machine called the CoffeePet attached with two small, OLED screen that will displays animated eyes. These eyes are responsive towards the user's gaze behavior. Furthermore, we used a sensor module (HVC Omron) to detect and track the eyes of a user in real time. It gives the ability for the user to interact with the CoffeePet simply by moving their eyes. The CoffeePet is also able to automatically brew and pour the coffee out of its spout if it feels appropriate during the interaction. We further explain the description of the system, modification of the real product, and the experimental plan to compare the user's perception of the CoffeePet's eyes and to investigate whether the user realizes or not that their gaze behavior influences the CoffeePet to react.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"281 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"115021988","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yosuke Fukuchi, Masahiko Osawa, H. Yamakawa, M. Imai
In cooperation, the workers must know how co-workers behave. However, an agent's policy, which is embedded in a statistical machine learning model, is hard to understand, and requires much time and knowledge to comprehend. Therefore, it is difficult for people to predict the behavior of machine learning robots, which makes Human Robot Cooperation challenging. In this paper, we propose Instruction-based Behavior Explanation (IBE), a method to explain an autonomous agent's future behavior. In IBE, an agent can autonomously acquire the expressions to explain its own behavior by reusing the instructions given by a human expert to accelerate the learning of the agent's policy. IBE also enables a developmental agent, whose policy may change during the cooperation, to explain its own behavior with sufficient time granularity.
{"title":"Autonomous Self-Explanation of Behavior for Interactive Reinforcement Learning Agents","authors":"Yosuke Fukuchi, Masahiko Osawa, H. Yamakawa, M. Imai","doi":"10.1145/3125739.3125746","DOIUrl":"https://doi.org/10.1145/3125739.3125746","url":null,"abstract":"In cooperation, the workers must know how co-workers behave. However, an agent's policy, which is embedded in a statistical machine learning model, is hard to understand, and requires much time and knowledge to comprehend. Therefore, it is difficult for people to predict the behavior of machine learning robots, which makes Human Robot Cooperation challenging. In this paper, we propose Instruction-based Behavior Explanation (IBE), a method to explain an autonomous agent's future behavior. In IBE, an agent can autonomously acquire the expressions to explain its own behavior by reusing the instructions given by a human expert to accelerate the learning of the agent's policy. IBE also enables a developmental agent, whose policy may change during the cooperation, to explain its own behavior with sufficient time granularity.","PeriodicalId":346669,"journal":{"name":"Proceedings of the 5th International Conference on Human Agent Interaction","volume":"6 1","pages":"0"},"PeriodicalIF":0.0,"publicationDate":"2017-10-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"122603962","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：