Kyutacar : Kyushu Institute of Technology Academic Repository
Not a member yet
9746 research outputs found
Sort by
時系列予測モデル・大規模言語モデルによる意味トリプルへの変換における自然言語記述の文法的複雑さの影響に関する研究
Get PDF九州工業大学博士（工学）1 Introduction| 2 Literature Review| 3 Basic Concept| 4 Methodology and Results| 5 Discussion and ConclusionIndustry 5.0 has put machines at the forefront in various industries. Machines are involved in every aspect of human life making it very critical to keep them in proper working condition. These machines are highly complex and rapidly evolving, resulting in a scarcity of highly skilled manpower capable of repairing them. One way to bridge this gap is to develop a knowledge-based system that can understand various machine components, their working, and causes of machine failure and thus help in the maintenance of machines and reduce the dependency on expert technicians. These knowledge bases or ontologies define concepts, relationships, and properties within a particular domain and can generate new inferences based on predicate logic. Manual generation of these concepts and relationships from raw text is very time-consuming and therefore in recent years machine learning techniques have been employed for these tasks, where a sentence is taken as an input and concepts and relations between them are extracted. For these knowledge-based systems to be dependable the concepts and relationship between them should be accurate.
The task of extracting concepts and relationships between them for ontology creation is called ontology population task. This Ontology Population Task (OPT) can be formulated as a classification task or a Neural Machine Translation (NMT) task. In the classification task, a sentence is given to a neural network model, and the model finds the words that belong either to a concept or a relation. In the case of NMT, an input sentence is translated to output a Resource Description Framework (RDF) triple. The current work aims to improve the quality of NMT task. The input sentences to a machine learning model can have different structures, the impact of these sentence structures on sequential model performance is not well studied. Most of the Natural Language Processing (NLP) applications are trained using annotated data without any importance given to the structure of the sentences used in training, this may lead to training data being skewed in terms of sentence structure and the distribution of sentences in training may differ from the distribution of sentences in real scenario. In this work to improve the quality of concepts and relations extracted from natural text, we analyze the effect of sentence structure on sequential models based on Bidirectional long short-term memory and Transformer architecture. We provide insight into the learning behavior of sequential models using statistical analysis methods like Kolmogorov-Smirnov test (KS test) and Cramer Von Mises test (CvM test). We also evaluate the model behavior on extraction task based on mean seeking forward Kullback-Leibler Divergence (KLD) and mode seeking backward KLD loss function.
Finally, the thesis contributes by providing mechanisms to improve the quality of concepts and relations extracted from natural text. The performance of the sequential model differs based on the loss function used for learning, a Modified Jeffreys Divergence (MJD) proposed in this work that combines the mean seeking behavior of forward KLD and mode seeking behavior of backward KLD contributes to the quality improvement. Based on the insight gained from the statistical analysis: the sequential model’s performance is affected by the structure of the sentences used for training and therefore the data used for training should have a proper distribution of different types of sentence structure, our proposal of a Structure Dependent Weighted Loss Function and the mechanism of selecting different model checkpoints based on sentence type also helped in improving the performance of the sequential model.九州工業大学博士学位論文 学位記番号:生工博甲第498号 学位授与年月日:令和6年9月25日令和6年度doctoral thesi
Towards Human-Level Evaluation: Assessing the Potential of GPT-4 in Automated Evaluation and Feedback Generation on Japanese Essays
Get PDFIn recent years, Automated Writing Evaluation (AWE) has been extensively researched within the field of AI in education. This paper explores generative AI, such as GPT-4, which has garnered significant attention for its ability to score essays and provide feedback to students. We designed prompts for GPT-4 to assign scores and rationales based on a given rubric and to generate feedback beneficial for students' development. We compared the evaluations produced by GPT-4 with those made by human evaluators. The results demonstrate GPT-4's potential to assist in generating evaluations at a human level. In addition, we analysed the consistency of the scoring and the quality of the rationales and feedback generated by GPT-4. In this paper, we will share our analysis and also describe the points that need to be improved for implementation in practice.journal articl
Design and Simulation of Piezoelectric Wideband Acoustic Sensor Covered with Organic Film
Get PDFA wideband acoustic sensor is reported in which the cantilevers are covered with an organic film in a piezoelectric micro-electro mechanical systems (MEMS) acoustic transducer. Coating the gap between the cantilevers with an organic film expands the low-frequency sensitivity, and connecting the cantilevers with an organic film increases the resonance frequency. Therefore, a broadband acoustic sensor can be realized [1]. However, coating with an organic film lowers the sensitivity, making the selection of the material for the organic film crucial. In this study, we perform finite element method (FEM) simulations using a combination of piezoelectric materials and organic films and provide guidelines for selecting organic films. Simulations are performed using AlN, ZnO, and polyvinylidene fluoride (PVDF) as piezoelectric materials. The results show that the larger the Young’s modulus ratio between the piezoelectric layer and the organic film, the lesser the effect of the organic film, which is expected to suppress the decrease in sensitivity caused by the organic film. The simulation results of the effective stress distribution from the fixed end to the free end of the cantilever confirms that there exists a region where the effective stress is high near the tip of the cantilever. This suggests that sensitivity can be improved by placing the electrode near the tip of the cantilever.journal articl
A Method of Improving the QOL of the People with Visual Impairment by MY VISION
Get PDFVisually impaired people face several difficulties in indoor activities, such as spending excessive time in locating objects. This paper proposes a method for assisting object acquisition by detecting desired objects and guiding users to them. The method requests a user to express the object he/she wants to acquire verbally and utilizes speech recognition to detect the specified object. Subsequently, the system guides in voice the user's hand to the location of the desired object. The performance of the method is experimentally shown. The method contributes to enhancing the comfort of indoor activities of visually impaired and, in this way, improves their quality of life.conference pape
A Method of Recognizing Body Movements Based on a Self-viewpoint Video
Get PDFThe most critical human sensory function resides in vision. This paper focuses on utilizing visual information, specifically self-perspective footage, to identify individual movements. Existing researchesrequire third-party filming to recognize human body movements and states. The proposed method, on the other hand, simply attaches a camera to the human head and enables the recognition of the subject's actions. Consequently, it becomes easier to monitor daily movements of a human and gather his/her data on body kinetics. This approach would be beneficial in scenarios involving individuals engaging in risky behavior or, during a certain emergency, providing valuable assistance.conference pape
Power-cycling degradation monitoring of an IGBT module with VCE(sat) measurement in continuous operation of a chopper circuit
Get PDFThis paper presents a power-cycling degradation monitoring method of an IGBT module with a VCE(sat) sensing circuit and junction temperature prediction by a three-dimensional structure model. A chopper circuit was introduced to provide a continuous-current-conducting operation of the IGBT module. The VCE(sat) sensing circuit with a low-cost IoT platform “Leafony” was utilized to monitor the junction temperature of an IGBT chip, which transferred the measured signal as digital data, and thus obtained a higher noise immunity than an analog-based circuit. The junction temperature of IGBT chip in the power module was analyzed from the dissipated power of IGBT and the transient thermal impedance between the chips and the ambient. This analysis is effective not only to observe the degradation but also to estimate the thermal resistance. Comparing the temperature profile between experiment and prediction provides health condition of the IGBT model.
Predicted thermal profiles agreed with measured ones with 10 % increase of thermal resistance, which was degraded by a power cycle tester. From these results, the proposed monitoring method is effective to detect the progress of power cycle degradation.journal articl
Relabeling for Indoor Localization Using Stationary Beacons in Nursing Care Facilities
Get PDFIn this study, we propose an augmentation method for machine learning based on relabeling data in caregiving and nursing staff indoor localization with Bluetooth Low Energy (BLE) technology. Indoor localization is used to monitor staff-to-patient assistance in caregiving and to gain insights into workload management. However, improving accuracy is challenging when there is a limited amount of data available for training. In this paper, we propose a data augmentation method to reuse the Received Signal Strength (RSS) from different beacons by relabeling to the locations with less samples, resolving data imbalance. Standard deviation and Kullback–Leibler divergence between minority and majority classes are used to measure signal pattern to find matching beacons to relabel. By matching beacons between classes, two variations of relabeling are implemented, specifically full and partial matching. The performance is evaluated using the real-world dataset we collected for five days in a nursing care facility installed with 25 BLE beacons. A Random Forest model is utilized for location recognition, and performance is compared using the weighted F1-score to account for class imbalance. By increasing the beacon data with our proposed relabeling method for data augmentation, we achieve a higher minority class F1-score compared to augmentation with Random Sampling, Synthetic Minority Oversampling Technique (SMOTE) and Adaptive Synthetic Sampling (ADASYN). Our proposed method utilizes collected beacon data by leveraging majority class samples. Full matching demonstrated a 6 to 8% improvement from the original baseline overall weighted F1-score.journal articl
Performance Evaluation of Spatio-temporal Data Retention System Supporting a Floating Cyber-physical System
Get PDFWe propose a floating cyber-physical system (F-CPS) as a novel platform for data distribution and application execution to realize a CPS. The F-CPS constructs a local-oriented and distributed CPS using devices with computing capability close to the user, such as smartphones, sensor devices, vehicles, and roadside units. This paper first provides an overview of the F-CPS and then discusses the challenges posed by the spatio-temporal retention system (STD-RS), a network technology for data distribution and maintenance in the local area of the F-CPS. To address the challenges associated the STD-RS, we propose adaptive data transmission control to effectively retain the STD in a certain area. Finally, we performed network simulations to demonstrate the effectiveness of the proposed adaptive method.journal articl
A Low-Area Overhead and Low-Delay Triple-Node-Upset Self-Recoverable Design Based on Stacked Transistors
Get PDFWith the aggressive scaling in the feature size of transistors, single-event triple-node-upsets (TNUs) induced by charge sharing in CMOS circuits have become a significant reliability problem. In this paper, based on N-type stacked transistors, a TNU self-recovery latch called LORD-TNU is proposed. Utilizing the stacked transistors to reduce the count of sensitive nodes in the latch. In addition, we use three modules to protect each other. In the event of a soft error in one module, the remaining modules can restore the corrupted module. This design not only saves delay overhead but also minimizes area overhead. Simulation results show that compared with the four typical TNU hardened latches, the proposed LORD-TNU latch reduces area overhead by 49.76%, power consumption by 56.07%, delay by 40.17%, and the power-delay-product (PDP) by 72.56% on average, respectively. Moreover, the robustness of our LORD-TNU latch is confirmed by comprehensive PVT (Process, Voltage, Temperature) and Monte Carlo simulations, demonstrating its stability across a range of process corners, supply voltage, and temperature variations.journal articl
Packet Recovery Method with Redundant Packets for Large Spatio-Temporal Data Retention
Get PDFCyber Physical System (CPS) is a system that accumulates information from physical space, analyzes them in cyberspace, and then feeds back the results. We aim to realize a Floating Cyber Physical System (F-CPS), a regionally distributed CPS. We have proposed a large-capacity spatio-temporal data retention system for delivering data and function (application) in the F-CPS. In a previous work, we proposed a data completeness-aware transmission control and evaluated it by simulation. However, considering the movement of obstacles and fluctuation of radio waves in the real environment, packet loss occurs frequently. Thereby, the system may not be able to have complete data and may not function. In this paper, we proposed a method to recover missing data using redundant packets and evaluated its impact on the entire system by testing it with actual equipment.journal articl