| 堀川 康生 | M, 1回目発表 | ソフトウェア設計学 | 飯田 元 | 井上 美智子 | 市川 昊平 | 柏 祐太郎 |
|
title: Toward Investigating the Impact of Test-Specific Refactoring
abstract: Refactoring refers to modifying a program without changing its external behavior to simplify its internal structure or improve its performance. Refactoring is performed not only on production code but also on test code. However, refactoring for test code has not been thoroughly studied, and the test refactoring patterns and their effects are yet to be revealed. In this study, we aim at identifying and classifying refactoring patterns specific to test code through a manual inspection. Additionally, we investigate how each identified test-specific-refactoring addresses potential issues in test code, known as test smells. language of the presentation: Japanese 発表題目: テストコード固有のリファクタリングが及ぼす影響調査にむけて 発表概要: リファクタリングは，内部構造の単純化や可読性の向上を目的として，外部から見た振る舞いを変更せずにプログラムを修正することを指す．その対象はプロダクションコードに留まらず，テストコードに対しても行われる．しかし，テストコードに対するリファクタリングは体系化されておらず，存在するリファクタリングのパターンやその効果が明らかになっていない．そこで，本研究では目視調査を通して，テストコード固有のリファクタリングパターンの特定・分類を行う．また，検出したテスト固有リファクタリングがテストコードにおける潜在的な問題（テストスメル）の解決にどの程度効果を有するかを調査する． | ||||||
| 渡邉 未来 | M, 1回目発表 | ソフトウェア設計学 | 飯田 元 | 松本 健一 | 市川 昊平 | 柏 祐太郎 |
|
title : Investigating the Impact of Shortening Release Cycle
on Introduction and Resolution of Self-Admitted Technical Debt
abstract: Technical Debt refers to the additional effort required in the future as a result of choosing an incomplete but immediate solution over an ideal implementation that would take more time. Specifically, when developers explicitly acknowledge technical debt in source code comments, it is referred to as Self-Admitted Technical Debt (SATD). In recent years, with the intensification of competition in software development, many projects have adopted practices to increase the frequency of major software releases. The acceleration of major releases can potentially facilitate the repayment of SATD by enabling developers to make flexible changes without concern for backward compatibility. On the other hand, shorter release intervals may increase psychological pressure on developers, potentially exacerbating the introduction of SATD. This study investigates how the frequency, resolution, and survival duration of SATD change before and after modifications in the release cycles, using Eclipse and Firefox as case studies. | ||||||
| 高橋 茉莉香 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 安本 慶一 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Ring-shaped device for converting visual information into tactile information
abstract: A tactile map is one method that enables visually impaired individuals to understand spatial information. While tactile maps can convey spatial information to users through tactile sensations, they have several challenges: the need for dedicated applications, limitations in media compatibility, and difficulties for sighted individuals to comprehend the tactile information. To address these issues, we propose a device that utilizes sensors to instantly convert visual information obtained from tablets or paper-based media into tactile information, aiming to overcome the aforementioned challenges. language of the presentation: Japanese 発表題目: 視覚情報を触覚情報へ変換するリング型デバイスの提案 発表概要: 目の不自由な人が空間を把握するための手段として，蝕地図がある．蝕地図は触覚情報を用いて空間情報をユーザに伝達できるが，専用のアプリケーションが必要である，媒体の制限が存在する，晴眼者が把握しづらいなどの課題が存在する．そこで，センサを用いてタブレットや紙媒体から得られる視覚情報を触覚情報に即座に変換するデバイスを提案し，前述の課題の解消を目指す． | |||||||
| 松﨑 優陽 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 向川 康博 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Proposal for an Analysis System to Support Efficient Improvement of Baseball Batting Form
abstract: In recent years, data collection in baseball has advanced significantly, enabling the measurement of metrics such as exit velocity and ball spin rate, which provide deeper insights into player performance. However, motion analysis and quantification remain underdeveloped, particularly for amateur players, where high costs and specialized equipment hinder widespread adoption. Existing systems also face challenges in user accessibility and practical application. To address these issues, this study proposes an affordable and user-friendly batting motion analysis system. By leveraging Generative Adversarial Imitation Learning, the system generates an ideal swing based on the user’s swing video and provides feedback in both visual and textual formats. This helps users improve their form efficiently. The system aims to reduce dependence on coaches and specific training environments, offering a new training method accessible to a wide range of players. language of the presentation: Japanese 発表題目: 野球における打撃フォームの効率的な改善を支援する解析システムの提案 発表概要: 近年、野球において打球速度やボールの回転数などのデータ化が進み、選手の能力を深く理解するための新たな指標が数多く生まれている。しかし、動作の分析や定量化はまだ発展途上であり、特にアマチュア選手向けの動作解析では、高コストや専用機器の必要性が普及の妨げとなっている。また、既存システムにはユーザーが活用しづらいといった課題も見られる。そこで本研究では、安価で誰でも使いやすい打撃動作解析システムを提案する。このシステムは、敵対的模倣学習を活用し、ユーザーのスイング動画を基に理想的なフォームを生成、映像とテキストでフィードバックを提供することで、効率的なフォーム改善を支援する。これにより、指導者や特定の練習環境への依存を減らし、新しい練習法を広く提供することを目指す。 | |||||||
| 松原 幹太 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 和田 隆広 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Effect of force haptic stimulation for the sole on redirected walking.
abstract: In virtual reality (VR), a key challenge for walking experiences arises when walking a vast VR space within the confines of a limited physical environment. To address this, Redirected Walking (RDW) is proposed, which creates an illusion of user movement and direction. However, if users become aware of RDW, their sense of immersion is compromised. To evaluate this, a metric known as the detection threshold is used to determine whether users notice the manipulation. Previous research has proposed using haptic feedback to expand this detection threshold. This study aims to investigate the detection threshold when force tactile feedback is applied to the soles of the feet. language of the presentation: Japanese 発表題目: 足裏力触覚刺激がRedirected Walkingに及ぼす効果に関する研究 発表概要: VRにおける歩行体験の課題として、広大なVR空間を歩行しようとした時に、実空間の広さが限られている。そこで、ユーザの移動量や移動歩行を錯覚させるRedirected walking（RDW）と呼ばれる手法がある。ユーザがRDWに気づくと没入感が損なわれてしまうため、気づいているか判断する検出しきい値という指標がある。この検出しきい値を拡大するために触覚提示を用いる手法がある。本研究では、足裏に力触覚提示した際の検出しきい値を調査することを目的とする。 | |||||||
| 宮本 杏樹 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 加藤 博一 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Vehicular telepresence with reduced communication delays by presenting images of the location and road surface conditions predicted from the input.
abstract: Telepresence refers to 'the feeling of being physically present in a remote or simulated location'.Teleoperation of robots with this telepresence is expected to improve efficiency and reduce the workload of operators.However, in an environment where communication delays occur during teleoperation, it becomes very difficult for the operator to operate the robot, leading to reduced work accuracy and increased task completion time.Therefore, we propose a system that enables stress-free operation in environments where communication delays occur, by reducing the perception of delay felt by the operator.As a proposed method, when an input is received from the operator, the video image at the predicted destination point is immediately presented to the operator from the input, before the robot moves to the point as input.The system also presents to the operator the predicted swaying of the robot based on the unevenness of the road surface at the destination, thereby strengthening the sense of physical and spatial unity, and aiming to create a system in which the operator does not feel any delay. language of the presentation: Japanese 発表題目: 入力から予測される地点の映像と路面状況の提示による通信遅延を軽減した車両型テレプレゼンス 発表概要: テレプレゼンスとは、「物理的に離れた場所やシミュレーションされた場所に存在する感覚」を指す。このテレプレゼンスを伴ったロボットの遠隔操作は、効率性の向上やオペレータの作業負荷の軽減が期待できる。しかし遠隔操作の際に通信遅延が生じる環境においては、オペレータにとって操作が非常に困難となり、作業精度の低下やタスク完了時間の増加に繋がる。よって、通信遅延が生じる環境でもオペレータが感じる遅延の知覚を軽減することで、ストレスを感じることなく操作が可能であるシステムを提案する。提案手法として、オペレータから入力を受け付けた際、ロボットが入力通りの地点へ移動するよりも前に、予測される移動先の地点での映像を入力から即時にオペレータに提示する。またこのとき、移動先での路面の凹凸から予測されるロボットの揺れも同時にオペレータに提示することで、身体的および空間的な一体感を強化し、オペレータに遅延を感じさせないシステムを目指す。 | |||||||
| 八島 永知 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 加藤 博一 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Development and Evaluation of a Spatial Information Transmission System Using Cheek-Based Tactile Feedback
abstract: Animals possess sense organs specialized for their respective species, allowing them to perceive their surroundings with remarkable precision. In contrast, humans primarily rely on vision and hearing, and when these senses are constrained, their spatial awareness tends to decline. In recent years, there has been growing interest in approaches that present spatial information through the sense of touch, thereby enhancing human spatial recognition ability. Tactile perception is distributed across the entire skin and enables the direct acquisition of information through physical contact. In this study, we propose a method for presenting spatial information using tactile feedback. Specifically, we construct a system that utilizes a LiDAR sensor to acquire the positions of surrounding objects, and then conveys this information through vibrational motors attached to the wearer’s cheeks. This system aims to facilitate effective spatial recognition even under conditions where vision and hearing are restricted. language of the presentation: Japanese 発表題目: 頬部触覚提示による空間情報伝達システムの開発と評価 発表概要: 動物は種ごとに特化した感覚器官を持ち，周囲の環境を精密に認識している．一方，人間は主に視覚や聴覚に依存しており，これらが制約される状況下では空間認識能力が低下するという課題がある．近年，触覚を通じて空間情報を提示し，人間の空間認識能力を向上させる試みが注目されている．触覚は皮膚全体に分布し，直接的な物理的接触を通じて情報を取得できる感覚である．本研究では，触覚を活用して空間情報を提示する手法を提案する．具体的には，LiDARセンサーを用いて取得した周囲の物体の位置情報を，頬に装着した振動モーターを通じて触覚的に提示するシステムを構築する．このシステムにより，視覚や聴覚に制約がある状況下でも，効果的な空間認識が可能になることを目指す． | |||||||
| 若山 陽登 | M, 1回目発表 | サイバネティクス・リアリティ工学 | 清川 清 | 加藤 博一 | 内山 英昭 | Perusquia Hernandez Monica | 平尾 悠太朗 |
|
title: Proposal of tactile integrated VR typing system
abstract: Text input in VR space has the problem that input actions cannot be performed smoothly due to the lack of tactile sensation and errors in estimating finger position and posture.Therefore, in this study, we introduce a text input system that integrates a tactile presentation device and investigate the accuracy and speed of text input in VR space by reducing the errors that occur in sensation and measurement. language of the presentation: Japanese 発表題目: 触覚統合型VRタイピングシステムの提案 発表概要: VR空間でのテキスト入力には，触覚の欠如や手指位置・姿勢推定の誤差によって入力動作が円滑に行えないという課題がある．そこで，本研究では触覚提示デバイスを統合したテキスト入力システムを導入し，感覚・計測において生じる誤差を低減することで，VR空間でのテキスト入力における精度と速度を調査する． | |||||||
| 坂上 温紀 | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | Sakriani Sakti | 上垣外 英剛 |
|
title: Dictionary-Based Cross-Lingual Vocabulary Transfers
abstract: When learning a new language, dictionaries are indispensable tools. While dictionaries are essential tools, regardless of their scale, they are also one of the materials created when describing a language. In other words, most documented languages have dictionaries. Looking at language models, when utilizing models trained on a high-resource language for a low-resource language, parallel corpora between them are used. However, such parallel corpora are challenging to obtain for some languages. Therefore, this study aims to transfer monolingual models between languages using bilingual dictionaries, which exist for many languages. language of the presentation: Japanese 発表題目: 辞書を用いた言語モデルにおける語彙の転移 発表概要: 新たな言語を学ぶとき、辞書は必要不可欠なツールである。辞書は重要なツールであると同時に、規模の大小はあれど、ある言語を記述しようとするときに作られる資料の一つでもある。つまり、記述がなされている言語の多くには辞書が存在しているといえる。言語モデルに目を向けると、高資源言語のデータで訓練されたモデルを低資源言語で活用しようとするときには高資源言語と低資源言語の対訳コーパスが用いられることがある。一方でそのような対訳コーパスが入手しにくい言語も少なくない。そこで、本研究では、多くの言語にある二言語間辞書を用いて単言語モデルを言語間で転移させることを目指す。 | |||||
| 吉田 大城 | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | Sakriani Sakti | 上垣外 英剛 |
|
title: Influence of Images on Language-only Tasks in Large-scale Vision Language Models
abstract: A Large-scale Vision Language Model (LVLM) is a language model capable of handling both images and language, enabling reasoning through their interaction. This shares commonalities with how humans process information. Humans, when presented with visual information in addition to scenarios that can be reasoned through language alone, may be influenced by that visual information, leading to conclusions or insights different from those derived from language alone. This phenomenon is known as priming. In this study, we examine the similarity between the behavior of LVLMs and humans from the perspective of priming. Specifically, we investigate whether inputting specific images affects the reasoning results when solving reasoning tasks that only require linguistic information using an LVLM. language of the presentation: Japanese 発表題目: 大規模視覚言語モデルにおける言語タスクへの画像の影響 発表概要: Large-scale Vision Language Model (LVLM) は、画像と言語の両方を扱うことが可能な言語モデルであり、これらの相互作用により推論を行う。これは、人間の情報処理の方法と共通する部分がある。人間は言語のみで推論可能な事象に対して、視覚情報が追加されることで、その視覚情報に惑わされた結果、言語のみの場合とは異なる結論や洞察が誘発されることがある。このような現象をプライミングと言う。本研究では、プライミングの観点から、LVLMと人間の振る舞いの類似性を検証する。具体的には、言語情報のみを必要とする推論タスクをLVLMで導く際に、特定の画像を入力することで、推論結果に影響を及ぼすかどうかを調査した。 | |||||
| RIERA MACHIN MARIA ANGELICA | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | 荒牧 英治 | 上垣外 英剛 |
|
title: Using the CEFR to Guide LLMs in Simplifying Spanish.
abstract: Lexical simplification involves modifying complex or difficult vocabulary and structures with simpler alternatives to make text more accessible while preserving its original meaning. This task is essential for several groups of people, including language learners. Large language models (LLMs) can play a significant role in this process due to their ability for generating simpler versions of text. Focusing on using the Common European Framework of Reference for Languages (CEFR) to guide LLMs in simplifying Spanish, a language with substantial resources but fewer compared to English, this study aims to align LLM text outputs with CEFR proficiency levels to improve Spanish language accessibility and support learners at different stages of language proficiency. language of the presentation: English | |||||
| 岩國 巧 | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | 荒牧 英治 | 上垣外 英剛 |
|
title: Combining metrics in machine translation evaluation for High-Risk Fields abstract: Reliable evaluation metrics are essential for translations in critical domains such as finance and medicine, where errors can have serious consequences. COMET is widely recognized for its high correlation with human evaluation, making it a popular choice for machine translation assessment. However, COMET does not account for variance, meaning high scores do not always reflect stable or reliable performance. To address this limitation, this study proposes a method that combines COMET with other evaluation metrics. By integrating multiple evaluation axes, this approach aims to capture aspects of translation quality that cannot be assessed by a single metric, enabling a more reliable and comprehensive evaluation for specialised domains. language of the presentation: Japanese | |||||
| 大熊 基暖 | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | 荒牧 英治 | 上垣外 英剛 |
|
title: A prompting method for response generation systems of educational dialogue for second-language acquisition
abstract: This study develops a dialogue system for providing educational interactions to second-language learners. Practical dialogue experience with a teacher plays a crucial role in second-language acquisition (SLA). While AI teachers effectively facilitate such dialogues, their performance still falls short of human capabilities with current technology. To address this, improvements are pursued by defining and instructing subtasks that enable multistage situational judgments, based on clearly specified requirements for the AI teacher. Designing these subtasks necessitates understanding real-world educational interactions, with an initial focus on dataset analysis. Although a zero-shot environment is assumed, comparisons with supervised learning models will be conducted to contribute new insights to the field of SLA. language of the presentation: Japanese | |||||
| 北野 雄士 | M, 1回目発表 | 自然言語処理学 | 渡辺 太郎 | 荒牧 英治 | 上垣外 英剛 |
|
title: Towards Interpretation in Multilingual Contextualized Embeddings with Independent Component Analysis
abstract: In natural language processing, embeddings can be either static or contextualized. Static embeddings, such as Word2Vec or GloVe, assign a fixed representation to each word, regardless of its usage. In contrast, contextualized embeddings, generated by models like BERT, adapt dynamically to the word's context in a sentence, enabling more nuanced understanding of language. This distinction is pivotal in tasks requiring precise interpretation of word meanings in varying contexts. However, these embeddings are incapable of being interpreted due to their abstract and high-dimensional representation. To address this issue, previous research has primary focused on interpretation in static embeddings with independent component analysis (ICA), while contextualized embedding in models effectively encode word meanings and their relationship. We apply ICA that transforms static embeddings into superior interpretable axes to contextualized embeddings for clarification of what pre-trained models learn. language of the presentation: Japanese | |||||