京都大学 / 0048 / 新制・課程博士 / 博士(情報学) / 甲第19342号 / 情博第594号 / 新制||情||103(附属図書館) / 32344 / 京都大学大学院情報学研究科システム科学専攻 / (主査)教授 石井 信, 教授 杉江 俊治, 教授 大塚 敏之 / 学位規則第4条第1項該当 / Doctor of Informatics / Kyoto University / DFAM

An imitation of a material is usually associated with a lower value. It could partly be because materials linked to permanence and aging are often seen as something more exclusive. In materials such as stone and wood, we can often see a trace of time. It can be seen through the growth rings in diverse types of wood. Or the pattern formed in rocks by magma that solidified in the earth’s crust over time. In my project, I want to question how we value materials and material imitations in relation to their age, execution, and location. By juxtaposing original materials with distorted copies, the materials communicate with each other and thus create an enhanced imitation. In my project, I investigate how we value materials, based on the idea of the material as something considered as genuine or fake and what is possible within the limitations of an imitation.

This study addresses the usefulness of elicited imitation (EI) and automated speech recognition (ASR) as a tool for second language acquisition (SLA) research by evaluating differences among learners of Japanese. The findings indicate that the EI and ASR grading system used in this study was able to differentiate between beginning- and advanced-level learners as well as instructed and self-instructed learners. No significant difference was found between self-instructed learners with and without post-mission instruction. The procedure, reliability and validity of the ASR-based computerized EI are discussed. Results and discussion will provide insights regarding different types of second language (L2) development, the effects of instruction, implications for teaching, as well as limitations of the EI and ASR grading system.

A principal argument against the use of elicited imitation (EI) to measure L2 oral proficiency is that performance does not require linguistic knowledge, but requires only rote memorization. This study addressed the issue by administering two tests to the same group of students studying English as a second language: (1) a working memory test, and (2) an English oral proficiency EI test. Participants came from a range of English language proficiency levels. A Pearson correlation was performed on the test results for each participant. The hypothesis was that English EI scores and working memory scores would not correlate significantly. This would suggest that the two tests do differ in what they measure, and that the English EI test does measure knowledge of the language to some degree. The results of the Pearson correlation revealed that there was a small positive correlation between working memory and English EI scores, but that it was not significant. There was also a significantly positive correlation between students' English EI scores and ELC level. These findings suggest that the English EI test fundamentally functions as a language test, and not significantly as a working memory test.

The objective and automatic grading of oral language tests has been the subject of significant research in recent years. Several obstacles lie in the way of achieving this goal. Recent work has suggested a testing technique called elicited imitation (EI) can be used to accurately approximate global oral proficiency. This testing methodology, however, does not incorporate some fundamental aspects of language such as fluency. Other work has suggested another testing technique, simulated speech (SS), as a supplement to EI that can provide automated fluency metrics. In this work, I investigate a combination of fluency features extracted for SS testing and EI test scores to more accurately predict oral language proficiency. I also investigate the role of EI as an oral language test, and the optimal method of extracting fluency features from SS sound files. Results demonstrate the ability of EI and SS to more effectively predict hand-scored SS test item scores. I finally discuss implications of this work for future automated oral testing scenarios.

Elicited Imitation (EI) is an assessment approach that uses sentence imitation tasks to gauge the oral proficiency level of test takers. EI tests have been created for several of the world's languages, including English, Spanish, Japanese, French, and Mandarin. Little research has been conducted for using the EI approach with learners of Russian. This dissertation describes a multi-faceted study that was presented in two journal articles for the creation and analysis of a Russian EI test. The EI test was created for and tested with Russian-speaking missionaries and employees at the Missionary Training Center (MTC) in Provo, UT. The first article describes the creation of the test and analyzes its ability to predict oral language proficiency by comparing individuals' scores on the EI to their scores on the Oral Proficiency Interview (OPI). The test was found to effectively predict an individual's OPI score (R2 = .86). The second article analyzes the difference in person ability estimates and item difficulty measures between items from a general content bank and a religious content bank. The mean score for the content specific items (x̄ = .51) was significantly higher than the mean score for the general test (x̄ = .44, p < 0.001). Additionally, the item difficulties for the religious items were significantly less than the item difficulties for the general items (p < 0.05).

The way characters move and behave in computer and video games are important factors in their believability, which has an impact on the player’s experience. This project explores Imitation Learning using limited amounts of data as an approach to creating human-like AI behaviour in games, and through a user study investigates what factors determine if a character is human-like, when observed through the characters first-person perspective. The idea is to create or shape AI behaviour by recording one's own actions. The implemented framework uses a Nearest Neighbour algorithm with a KD-tree as the policy which maps a state to an action. Results showed that the chosen approach was able to create human-like AI behaviour while respecting the performance constraints of a modern 3D game. / Sättet karaktärer rör sig och beter sig på i dator- och tvspel är viktiga faktoreri deras trovärdighet, som i sin tur har en inverkan på spelarens upplevelse. Det här projektet utforskar Imitation Learning med begränsad mängd data som etttillvägagångssätt för att skapa människolik rörelse för AI-karaktärer i spel, ochutforskar genom en användarstudie vilka faktorer som avgör om en karaktärär människolik, när karaktären observeras genom dess förstapersonsperspektiv. Iden är att skapa eller forma AI-beteende genom att spela in sina egna handlingar. Det implementerade ramverket använder en Nearest Neighbour-algoritmmed ett KD-tree som den policy som kopplar ett tillstånd till en handling. Resultatenvisade att det valda tillvägagångssättet lyckades skapa människolikt AI-beteende samtidigt som det respekterar beräkningskomplexitetsrestriktionersom ett modernt 3D-spel har.

Stilkopien anzufertigen ist ein zentraler Bestandteil von Satztechnik- bzw. Tonsatzseminaren. Ausgehend von der Komposition Erinnerung aus dem Album für die Jugend op. 68 von Robert Schumann werden hier die Arbeitsschritte von analytischen Vorüberlegungen bis hin zum fertigen Klaviersatz für ein lyrisches Klavierstück präsentiert, Strategien für verschiedene Entwürfe vorgestellt und deren Ergebnisse diskutiert. Dabei kommen melodische, harmonisch-metrische und formale Kriterien ins Spiel. Für Begleitfiguren gibt es einen Exkurs zu Mendelssohns Lieder ohne Worte. / The production of style-imitations is central aspect of compositional seminars. Proceeding from the work Erinnerung from Schumann’s Album für die Jugend op. 68, this article presents the steps ranging from preliminary analytical considerations up to the completion of the keyboard score for a lyrical piece, while strategies for various drafts are offered and their results are discussed. Melodic, harmonic-metrical, and formal criteria all play a role. Regarding accompanimental figuration, a digression addresses Mendelssohn’s Lieder ohne Worte.

Cette thèse interroge la manière dont les penseurs des Lumières françaises ont compris, analysé et fondé philosophiquement les effets de sens de la musique, c’est à dire aussi bien les modalités selon lesquelles la musique produit du sens, que les modalités selon lesquelles l’être humain est affecté par la musique. Elle interroge plus généralement la place qu’occupe la pensée musicale dans l’économie des savoirs sur l’être humain au siècle des Lumières. Le sujet de la musique y est envisagé sous trois angles : comme corps physique soumis à un phénomène acoustiquement structuré (partie I), comme être parlant confronté au phénomène expressif primordial de la voix (partie II) et enfin comme imagination confrontée à des fictions sonore ou des imitations (partie III). / This PHD thesis examines how French philosophers from the enlightenment understood and analyzed musical effects, that is, how music produces meaning, but also, how the human being is affected by music. It is more generally about the specific place held by musical thought in the global anthropologic knowledge of the eighteenth century. The musical subject is here considered in three manners: first, as a body affected by an acoustically structured phenomenon, second, as a speaking animal affected by voice, and finally as a mind confronted to musical fictions or imitations.

<p> </p> <p>The usage of robots has increased in different areas of society and human work, including medicine, transportation, education, space exploration, and the service industry. This phenomenon has generated a sudden enthusiasm to develop more intelligent robots that are better equipped to perform tasks in a manner that is equivalently good as those completed by humans. Such jobs require human involvement as operators or teammates since robots struggle with automation in everyday settings. Soon, the role of humans will be far beyond users or stakeholders and include those responsible for training such robots. A popular teaching form is to allow robots to mimic human behavior. This method is intuitive and natural and does not require specialized knowledge of robotics. While there are other methods for robots to complete tasks effectively, collaborative tasks require mutual understanding and coordination that is best achieved by mimicking human motion. This mimicking problem has been tackled through skill imitation, which reproduces human-like motion during a task shown by a trainer. Skill imitation builds on faithfully replicating the human pose and requires two steps. In the first step, an expert's demonstration is captured and pre-processed, and motion features are obtained; in the second step, a learning algorithm is used to optimize for the task. The learning algorithms are often paired with traditional control systems to transfer the demonstration to the robot successfully. However, this methodology currently faces a generalization issue as most solutions are formulated for specific robots or tasks. The lack of generalization presents a problem, especially as the frequency at which robots are replaced and improved in collaborative environments is much higher than in traditional manufacturing. Like humans, we expect robots to have more than one skill and the same skills to be completed by more than one type of robot. Thus, we address this issue by proposing a human motion imitation framework that can be efficiently computed and generalized for different kinematic structures (e.g., different robots).</p> <p> </p> <p>This framework is developed by training an algorithm to augment collaborative demonstrations, facilitating the generalization to unseen scenarios. Later, we create a model for pose imitation that converts human motion to a flexible constraint space. This space can be directly mapped to different kinematic structures by specifying a correspondence between the main human joints (i.e., shoulder, elbow, wrist) and robot joints. This model permits having an unlimited number of robotic links between two assigned human joints, allowing different robots to mimic the demonstrated task and human pose. Finally, we incorporate the constraint model into a reward that informs a Reinforcement Learning algorithm during optimization. We tested the proposed methodology in different collaborative scenarios. Thereafter, we assessed the task success rate, pose imitation accuracy, the occlusion that the robot produces in the environment, the number of collisions, and finally, the learning efficiency of the algorithm.</p> <p> </p> <p>The results show that the proposed framework creates effective collaboration in different robots and tasks.</p>