Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
Synthesis and Characterization of Multifunctional Symmetrical Squaraine Dyes for Molecular Photovoltaics by Terminal Alkyl Chain Modifications
Get PDFNovel far-red sensitive symmetric squaraine (SQ) dyes with terminal alkyl chain modifications were designed, synthesized, and characterized, aiming towards imparting multifunctionalities such as photosensitization, dye aggregation prevention, and source of electrolyte components. The dye sensitizer SQ-80 with alkyl chain terminal modifications consisting of 1-methylimidazolium iodide was designed and synthesized as a new dye sensitizer for DSSCs based on symmetric SQ-4 without any terminal modification used as reference. Upon adsorption on the mesoporous TiO2 surface, SQ-80 demonstrated reduced dye aggregation and stronger binding to the TiO2 surface, leading to enhanced durability of DSSCs. Apart from the most common photosensitization behavior, the newly designed dye demonstrated multifunctionalities such as aggregation prevention and electrolyte functionality, utilizing iodine-based redox electrolytes in the presence and absence of I2 and LiI additives. In the absence of LiI and I2, a mixture of SQ-77 with alkyl chain terminal modifications consisting of iodide and SQ-80 demonstrated a photoconversion efficiency of 1.54% under simulated solar irradiation, which was about six times higher compared with the reference dye SQ-4 (0.24%) (having no alkyl chain terminal modification).journal articl
Personalized assist-as-needed dressing assistance robot without human modeling using Rowat-Selverston CPG controller
Get PDFDressing is a critical component of Activity of Daily Living (ADLs). The development of assistive technology for dressing tasks is urgently needed from the viewpoint of privacy, for example, to prevent the wearer from being seen in the nude. Despite their importance, these technologies are underutilized in garment donning compared to other ADLs. The reasons for the lack of utilization of assistive technology are that the task is complex, as it involves handling individuals with large individual differences in height and symptoms and requires manipulation of flexible clothing. Addressing these variances presents a formidable academic challenge. Our previous study identified periodic patterns in human movements during robot-assisted dressing and noted individual variability in these patterns. Capitalizing on this finding, we propose a novel control strategy for a robot that adapts to individual needs during dressing, employing the ‘Assist-As-Needed’ (AAN) principle from physical therapy. The proposed control method uses Central Pattern Generators (CPG) that can be synchronously controlled in response to external forces, enabling model-free control without human modeling. We utilize the Rowat-Selverston CPG model, which is recognized for its adaptive response to human motions in human-robot interactions such as with handshaking robots. A simulator of the Rowat-Selverston CPG was created. The output of the CPG was confirmed by inputting the data set obtained in previous studies, and the parameters were adjusted. The CPG output was then applied to the recorded target joint trajectory for the dressing assistance. We prepared a replay of the default trajectory, a trajectory with simple harmonic motion applied, and a trajectory with CPG output applied, and confirmed whether individual adaptation according to the AAN was possible through subject experiments. The experimental results showed that the Rowat-Selverston CPG control method enables individual adaptive dressing assistance according to the AAN principle. This study summarizes these methods and results and contributes to the realization of an individually adaptive system that follows the AAN principle, especially in developing assistive technology.journal articl
Bidirectional 2D reservoir computing for image anomaly detection without any training
Get PDFImage anomaly detection is a crucial task in computer vision, where convolutional neural networks (CNN) often deliver exceptional performances. Hardware implementation of machine learning models is also important for achieving inference speed-up and power savings. However, the massive number of CNN parameters poses challenges for hardware implementation. This study introduces reservoir computing (RC) to create a compact image processor without training, thereby enabling scalable deployment. Our proposed bidirectional 2-dimensional reservoir computing (BiRC2D) is a feature extractor based on RC. Experiments conducted on the MVTec AD dataset, a benchmark dataset for real-world anomaly detection task, validated the efficacy of BiRC2D when integrated into the patch distribution modeling (PaDiM) framework. The mean intersection over union (mIoU) score from PaDiM with BiRC2D outperformed or was comparable to the mIoU score from PaDiM with ResNet-50 in several categories while reducing the parameter count by up to 98%.journal articl
Analysis of Human Activity Recognition by Diffusion Models
Get PDFThis paper proposes a classifier based on a diffusion model for human activity recognition. To this end, we introduce three architectures: (1) the representation-conditional diffusion transformer, (2) the first classifier, RepcondFormer, and (3) the second classifier, RepcondClassifier. Experimental results show that RepcondFormer outperforms in three datasets, while RepcondClassifier performs better in one dataset when fine-tuned.journal articl
A Quality Analysis of the BiLSTM Encoder Decoder Model with Modified J-Divergence for Sentences with Different Complexities
Get PDFThe rapid increase in the field of artificial intelligence has made machines capable of making intelligent decisions. Machines can now recognize patterns and support human intelligent activities, and one of the main sources for machines to learn is natural languages. Natural languages are complex and sentences can have different structures, difficult for a machine to understand. A formal way to represent language is ontology, and ontology organizes information as a triple and set of rules. The task of extracting triples from various sources such as text documents, or web pages is termed as ontology population task that can be formulated either as a classification or a translation task. In both these tasks data from an unknown probability distribution is fed to a neural network that generates another probability distribution parametrized by network weight, and the objective here is to minimize the distance between two probability distributions by minimizing some loss function. In this research, we proposed a modified version of Jeffreys divergence as a learning method for machines to improve the ontology population task. We also analyze the performance of the neural network model based on different structures of sentences used for training and propose data selection method for improving model performance.journal articl
Self-assembling into nanostructure of Cu-Ag NPs with both enhanced antimicrobial activity and cytocompatibility
No full textThe germ-killing functions of silver and copper nanoparticles (NPs) were worthy of biomedical applications, but as heavy metals, NPs need to be more antimicrobial at lower doses to minimize hazards to human health, which requires more explorations on the new structures resulting from the compositing between two metals. In this study, self-assembled Cu-Ag bimetallic NPs were synthesized using a two-step reduction method. Results showed that both the Cu and Ag in Cu-Ag NPs existed as pure phases, the nanoparticles comprised ∼5 nm primary particles that self-assembled into ∼50 nm secondary particles surrounded with carbon-chain coatings. The larger secondary particles provided Cu-Ag with good particle dispersibility and chemical stability against deeper oxidation above ∼360 °C, while the smaller primary particles facilitated their suspensions in achieving a maximum inhibition ring of 10.01±0.25 mm and a minimum optical density (OD) of 0.11±0.01. Cu-Ag NPs demonstrated superior antimicrobial performance against all target strains, exhibiting synergistic enhancement through released active ions and reactive oxygen species (ROS). Cell viability ranged from 74.4 % to 95.4 % after co-culturing for 24 h, confirming that all synthesized nanometals were non-toxic to L929 cells at most concentrations. Increasing the suspension concentration enhanced antimicrobial activity while reducing cell viability. This conflict established an optimal Cu/Ag ratio of 10:4 and a concentration of 4 μg/mL, allowing Cu-Ag NPs to perform both broad-spectrum bactericidal activity and cell safety.journal articl
Анализ электромагнитного поля и повышение производительности модели сверхпроводящей полировальной машины с использованием ребер сверхпроводящей ленты
Get PDFIn this study, we have focused on superconductive-assisted polishing machine that is mainly composed of superconducting bulks and permanent magnet. This machine utilizes magnetic levitation, which enables processing in mid-air and inside a hollow object. In our previous work, we used the superconducting tapes instead of superconducting bulks for improvement, and confirmed that the performance was well as using superconducting bulks. Therefore, we considered that it is necessary to evaluate the performance of the superconductive-assisted polishing machine using superconducting tapes. We compared the previous model and the model with ribs which had a better result in the experiment. It is expected that the ribs will trap the magnetic flux so that we can get greater repulsive force. However, ribs did not work well enough in the calculations result using finite element method, because the magnetic flux did not penetrate vertically into the ribs.Статья посвящена исследованию полировальной машины со сверхпроводящим приводом, состоящим из сверхпроводящих материалов и постоянного магнита. В этой машине использована магнитная левитация, позволяющая проводить обработку в воздухе и внутри полого объекта. В предыдущем исследовании вместо объемных сверхпроводников для упрощения конструкции были использованы сверхпроводящие ленты. Результаты исследования показали, что производительность машины одинакова как при использовании сверхпроводящей ленты, так и объемных сверхпроводников. Поэтому оценка эффективности полировальной машины со сверхпроводящим приводом проводилась при использовании сверхпроводящих лент. В эксперименте проведено сравнение двух моделей с сверхпроводящими деталями: без ребер жесткости и с ребрами жесткости. Последняя показала лучший результат. Ожидается, что ребра позволят удерживать магнитный поток, благодаря чему будет возможно получить большую силу отталкивания. Однако расчет методом конечных элементов показал обратное: ребра работали недостаточно хорошо, поскольку магнитный поток не проникал вертикально внутрь ребер.journal articl
Addressing class imbalance in customer review: Analysis using focal loss and SVM with BERT
Get PDFIn today's digital marketplace, customer reviews play a critical role in influencing consumer decisions and in-forming business improvements. Among these, Request" and "Complaint" reviews provide direct insights into customer needs and areas of dissatisfaction. However, they often constitute a minority in review datasets, creating a class imbalance problem that hinders effective classification. In our research, we propose a novel approach to addressing class imbalance by incorporating Focal Loss into the fine-tuning of a BERT model for classifying customer reviews. Using a dataset with "Request", "Complaint", and other comment types, we demonstrate that Focal Loss significantly improves classification for the highly underrepresented "Request" class. Additionally, replacing BERT's fully connected layer with an SVM classifier further enhances performance on the "Request" class. However, we observed a slight decrease in classification effectiveness for the "Complaint" class, suggesting that complementary techniques may be necessary to achieve balanced performance. Our approach offers a robust solution for enhancing customer review analysis, enabling businesses to better capture and respond to critical customer insights."journal articl
Analysis of Factors Contributing to Missed Detections of Vehicles with Location Information Errors in STD-RS
Get PDFWe have proposed an STD-RS [1] system that uses vehicles for spatio-temporal data (STD) distribution to achieve local consumption of IoT data. However, in STD-RS, data distribution is based on the vehicle's own location information, which can lead to unnecessary data distribution due to location information errors. To address this issue, in our prior work [2], we proposed a method that uses machine learning with the average and standard deviation of RSSI (Received Signal Strength Indicator) concerning distance as features to detect vehicles with location errors. Although the proposed method achieved 80% detection accuracy, some vehicles were missed due to the large amount of data required for detection. Therefore, in this study, we evaluate the feature generation status of vehicles in the environment using simulations.conference pape
Data Transfer Control Method of Indoor Spatio-Temporal Data Retention in F-CPS
Get PDFA Floating Cyber Physical System (F-CPS) has been proposed as a region specfic Cyber Physical System (CPS). F-CPS promotes flexible data utilization by retaining Spatio-Temporal Data (STD), which depends on the location and time of data generation, within a specific area. We have proposed a Spatio-Temporal Data Retention System (STD-RS) to deliver STD in the F-CPS. In our previous studies, we assumed that nodes controlled data transmission based on GPS information in outdoor environments. However, F-CPS is also assumed for data retention in indoor environments. In this study, we propose a data transfer control method based on the received signal strength indicator to check its location information without using GPS. Additionally, we conducted experiments in a real environment and demonstrated that the proposed method effectively operates as an STD-RS in indoor environment.journal articl