Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Effect of R-phase on shape recovery speed of Ti-Ni shape memory alloy wire for variable-stiffness mechanism using jamming transition phenomenon
Get PDFThe jamming transition phenomenon refers to the change in stiffness of a powder as its density changes. As a familiar example, vacuum-packed coffee grounds are hard, but when opened to allow air into the package and the density is reduced, the grounds become softer. Using Ti-Ni shape memory alloys (SMAs) and the jamming transition phenomenon, a variable-stiffness and deformable link was devised and prototyped by Takashima et al. (2022). Although the mechanism worked well, it operated slowly due to the slow shape recovery velocity of the SMA. For practical application of this mechanism, it is necessary to improve the recovery velocity of Ti-Ni SMA. Normally, Ti-Ni SMAs exhibit an austenite to martensite phase transformation; however, Miyazaki and Otsuka (1984) observed that under some conditions, R-phase transformation can occur between these phase transformations. In a previous study, Tanikata et al. (2022) observed that SMAs that show R-phase transformation during shape recovery tend to have higher recovery stress than SMAs that do not show R-phase transformation. Therefore, SMAs that show R-phase transformation during shape recovery are expected to have a high shape recovery velocity. This study investigates the effect of R-phase transformation on the recovery velocity of SMA. From the experimental results, specimens that show R-phase transformation tend to have a higher recovery velocity than specimens that do not show R-phase transformation. However, the mechanism created using materials showing R-phase transformation had a small amount of shape recovery during shape recovery. Thus, we attempted to improve the mechanical properties of the SMA by training. The device using the trained specimen demonstrated increased shape recovery compared to the device using the untrained specimen.journal articl
Improved Evaluation Metrics for Sentence Suggestions in Nursing and Elderly Care Record Applications
Get PDFThis paper presents a new approach called EmbedHDP, which aims to enhance the evaluation models utilized for assessing sentence suggestions in nursing care record applications. The primary objective is to determine the alignment of the proposed evaluation metric with human evaluators who are caregivers. It is crucial due to the direct relevance of the provided provided to the health or condition of the elderly. The motivation for this proposal arises from challenges observed in previous models. Our analysis examines the mechanisms of current evaluation metrics such as BERTScore, cosine similarity, ROUGE, and BLEU to achieve reliable metrics evaluation. Several limitations were identified. In some cases, BERTScore encountered difficulties in effectively evaluating the nursing care record domain and consistently providing quality assessments of generated sentence suggestions above 60%. Cosine similarity is a widely used method, but it has limitations regarding word order. This can lead to potential misjudgments of semantic differences within similar word sets. Another technique, ROUGE, relies on lexical overlap but tends to ignore semantic accuracy. Additionally, while BLEU is helpful, it may not fully capture semantic coherence in its evaluations. After calculating the correlation coefficient, it was found that EmbedHDP is effective in evaluating nurse care records due to its ability to handle a variety of sentence structures and medical terminology, providing differentiated and contextually relevant assessments. Additionally, this research used a dataset comprising 320 pairs of sentences with correspondingly equivalent lengths. The results revealed that EmbedHDP outperformed other evaluation models, achieving a coefficient score of 61%, followed by cosine similarity, with a score of 59%, and BERTScore, with 58%. This shows the effectiveness of our proposed approach in improving the evaluation of sentence suggestions in nursing care record applications.journal articl
Toward Extracting Care Records from Transcriptions of Visiting Nurses Using Large Language Models
Get PDFIn this paper, we provide prompts to extract values or short texts for body temperature, appetite, daily activities, and mental state from automated transcripts of nurses’ words. However, because the conversation is intended to be as relaxed as possible, there are challenges in summarizing these dialogues into specific information that matches the care records. For example, some information in the transcript does not clearly convey certain information. In response to these challenges, we propose using Large Language Models (LLMs) to extract transcript information related to care records automatically. This paper investigates the effectiveness of LLMs in automatically identifying and extracting relevant care record information from communication sessions through automated transcription, with the ultimate goal of simplifying the documentation process and improving the quality of services provided to the elderly population. As a result, the prompts we generate for compact output types, such as body temperature and appetite, produce outputs that closely match the ground truth data. In thirteen documents tested with this prompt, eight of them both body temperature and appetite, produced information that was in line with ground truth.journal articl
Performance variation caused by sprite drawing pattern for high-level synthesized sprite drawing hardware
Get PDFWe are developing a sprite drawing software suitable for automatic hardware generation by high-level synthesis. Sprite drawing stores the sprite image into a frame memory pointed to by the origin coordinates. Frame memory is typically realized with high-capacity DRAM such as DDR SDRAM. The DDR SDRAM is prone to stalls when the memory accesses have randomness. The performance of sprite drawing hardware may depend on the drawing pattern on the frame memory in contrast with a CPU hiding the frame memory by a cache memory. This paper investigates how hardware and software performance varies with varying sprite drawing patterns. The results show that hardware achieves better real-time performance than software execution, although the performance variation of hardware due to drawing conditions is slightly larger than that of software execution.journal articl
Centrifuge Model Tests on Efeect of Non-Woven Fabrics in Concrete-Covered Levees on Deformation Process at High Water Levels
Get PDF本研究では，表面被覆型の堤防強化における被覆コンクリート背面の不織布の有無が高水位作用時の堤体変状プロセスに与える影響について，遠心模型実験を用いて基礎的な検討を行った．不織布有りのケースでは，川裏法尻付近から河川水が噴出し，法留工の傾斜とコンクリートブロックの滑動は確認できたものの，明確な堤体の侵食は確認できなかった．一方で，不織布無しのケースでは堤内地盤へ転倒した法留工とコンクリートブロックの間に発生した隙間から大量の濁水が噴出するとともに，コンクリートブロック背面の堤体が侵食された．また，本実験結果は過去に実施された海岸堤防に対する実大越流実験結果と整合しており，本遠心模型実験においても実堤防で発生する主要な現象は再現できていることを確認した．journal articl
SDNのプログラマブルデータプレーンによるネットワーク主導型TCPフロー制御調和手法
Get PDFSoftware Defi ned Networking（SDN）技術の一つであるプログラマブルデータプレーンによってネットワーク内で任意のパケット処理を実装できるようになった．ネットワーク内では，一つの端末では得られない，競合する通信フローの情報が容易に把握できるため，プログラマブルデータプレーンでその情報を活用して任意の新しいパケット制御を実現できれば，端末のみでは難しかった競合フローを直接的に意識した通信制御の実現が期待できる．本研究では，アプリ要求を意識したTCPフローの競合制御に着目する．一般的なTCPでは，競合フローとのスループットの公平性のみを目指すが，実際にはアプリ要求との不一致が生じる．つまり，競合フロー間でスループットが公平であっても，一方のアプリ要求には過剰，若しくは不足する，といった非効率な資源利用状態に陥る．この不一致解消には，アプリ要求と競合関係を理解し，スループットに意図的な傾斜を付ける制御が必要となる．本稿では，プログラマブルデータプレーンとその開発言語の P4 を解説し，上記の制御の実現に向けた初期検討として，競合する2フローのTCPスループットを事前に定義したアプリ要求に合わせて調和させるTCPフロー制御調和手法を試作する．journal articl
An ATP detection system based on the enzyme reaction with biotin protein ligase
No full textAdenosine triphosphate (ATP) is the energy currency of all living organisms and can be used as an indicator for cell proliferation and cytotoxicity. In the present work, we have developed a novel ATP detection system by combining the biotinylation reaction from archaeon Sulfolobus tokodaii with fluorescence resonance energy transfer (FRET). In biotinylation from S. tokodaii, an enzyme known as biotin protein ligase (BPL) forms a very stable complex with its product, biotinylated substrate protein (BCCP). Here, BPL and BCCP were fused to the fluorescent proteins Cerulean and Clover, respectively, and ATP detection was accomplished by monitoring the FRET signal between the two fluorescent proteins, since ATP is an essential component for biotinylation and the tight BPL-BCCP complex is formed only after biotinylation. Using this system, we have succeeded in detecting 5 nM of ATP by biotinylation reaction with 50 nM of each fusion protein. Our method has a characteristic that the signal does not decay for at least 2 h after the start of the reaction, unlike in the case of the luminescence-based assay with luciferase commonly used for the ATP detection. Thus, our system allows for ATP detection which is not significantly constrained by measurement timing.journal articl
Illust Protection against Generative AI using DnCNN
Get PDFStable Diffusionをはじめとする画像生成AIは急速に発展しているが，その一方でクリエイターの画像作品を無許可で模倣するなどの問題が懸念されている．この問題に対して，画像作品に対して細かな摂動を加えることで画像生成AIによる模倣を防止する手法等が提案されている．しかし，これらの手法は画像１枚毎に最適化が必要であり，計算コストが高い．本研究では，摂動付与を深層学習モデルで代替することで，従来の最適化を省略し，高速に摂動を付与する手法を提案する．実験の結果，提案手法の摂動生成速度は従来手法と比べて高速であった．また，摂動の画像保護性能についても従来手法と同程度であるのを確認した．Although generative AI such as stable diffusion are rapidly developing, but there are concerns about problems such as unauthorised mimicking of creators’ illustrations. In order to address this problem, some researchers proposed methods such as adding small perturbations to the illustrations for prevention Generative AI from mimicking the illustration. However, these methods require a high computational cost due to the complex optimization. In this study, we propose a novel method to replace the perturbation assignment with a deep learning, which eliminates the optimization step and significantly accelerate conventional methods. As a result of experiments, the proposed method runs significantly faster than conventional methods. Our image protection performance of the perturbation was also confirmed to be comparable to the previous method.journal articl