13197 research outputs found
Sort by
Toward fast meeting transcription: NAIST system for CHiME-8 NOTSOFAR-1 task and its analysis
No full textThis paper reports on the NAIST system submitted to the CHIME-8 challenge’s NOTSOFAR-1 (Natural Office Talkers in Settings of Far-field Audio Recordings) task, including results and analyses from several additional experiments. While fast processing is crucial for real-world applications, the CHIME-7 challenge focused solely on reducing error rate, neglecting the practical aspects of system performance such as inference speed. Therefore, this research aims to develop a practical system by improving recognition accuracy while simultaneously reducing inference speed. To address this challenge, we propose enhancing the baseline module architecture by modifying both the CSS and ASR modules. Specifically, the ASR module was built based on a WavLM large feature extractor and a Zipformer transducer. Furthermore, we employed reverberation removal using block-wise weighted prediction error (WPE) as preprocessing for the speech separation module. The proposed system achieved a relative reduction in tcpWER of 11.6% for single-channel tracks and 18.7% for multi-channel tracks compared to the baseline system. Moreover, the proposed system operates up to six times faster than the baseline system while achieving superior tcpWER results. We also report on the observed changes in system performance due to variations in the amount of training data for the ASR model, as well the impact of the maximum word-length setting in the transducer-based ASR module on the subsequent diarization system, based on findings from our system development.journal articl
Addressing Variations in Behavioral Biometrics for Continuous Authentication in Internet Banking on Desktop and Laptop Devices
Get PDF奈良先端科学技術大学院大学修士（工学）master thesi
Construction of novel organic supramolecular polymers and their integration with CsPbBr3 perovskite quantum dots
Get PDF奈良先端科学技術大学院大学博士（理学）doctoral thesi
A New Fault Detection Method Using Machine Learning in Analog Radio-on-Fiber MIMO Transmission System
No full textIn this letter, we propose a new fault detection method in a wired-wireless multiple-input and multiple-output (MIMO) transmission over an analog radio-on-fiber (A-RoF) network. The A-RoF architecture makes the configuration of the remote node simple, however, lots of radio frequency (RF)-band analog devices are used in it, such as semiconductor fiber laser, optical fiber link, photo-detector and electrical amplifiers, and the methodology for detecting the fault on these devices has not been discussed. Conventional methods require dedicated devices for sending fault messages via out-of-band wireless data transmission, which increases the hardware requirement of the whole system configuration. This letter proposes a machine learning approach in an A-RoF-based MIMO transmission without any additional devices.journal articl
Monitoring Over-The-Counter Drug Misuse in Japanese User-Generated Data
No full textIntroduction: The misuse of over-the-counter (OTC) drugs poses a significant global public health challenge. This study proposes a system for detecting and visualizing inappropriate OTC drug use in social media data. Methods: We constructed a corpus of 20,036 labeled Japanese tweets, including 7,000 medication-related posts, to address the linguistic and cultural nuances. By fine-tuning the Japanese bidirectional encoder representations from transformers models, the system identified misuse patterns such as overuse. The system also incorporates a visualization tool to illustrate temporal and categorical trends, aiding public health authorities in real-time pharmacovigilance efforts. Results: The system demonstrated a strong performance in detecting specific misuse patterns and has the potential to provide insights through the visualization of temporal and categorical trends. Error analysis revealed challenges such as ambiguous terms and noise inherent in social media data. Discussion: The model performed well with sufficient data, but struggled with underrepresented categories. Challenges with ambiguous terms and indirect references emphasize the need for improved contextual understanding and the potential benefits of larger language models or data augmentation techniques. Conclusion: Although this study focused on the Japanese context, the system identified OTC drug misuse patterns and provided information through visualization. This holds promise for real-time pharmacovigilance and can be applied to other languages, contributing to global efforts to monitor and mitigate drug misuse trends.conference pape
The filopodia dynamics via actin remodeling mediated by unconventional myosin and actin severing cofilin
Get PDF奈良先端科学技術大学院大学博士（バイオサイエンス）doctoral thesi
Efficient IDS for IoT Networks Using Host-Based Data Aggregation and Multi-Entropy Analysis
No full textT IoT devices have limited computational resources, posing challenges to implementing adequate security measures. As a result, numerous attacks targeting vulnerabilities in IoT devices have been observed. Against this backdrop, research on Intrusion Detection Systems (IDSs) leveraging machine learning in IoT environments has been actively conducted. However, packet-based and flow-based IDSs proposed in existing studies are vulnerable to attacks such as DoS and DDoS, which involve numerous packet or flow combination patterns. These methods also face challenges related to computational resource burdens caused by the increased volume of input data. This study proposes a lightweight IDS with the hostbased approach, representing communication behaviors with multiple entropies. The host-based approach aggregates features from different communications sent by the same host, enabling a reduction in input data. Additionally, the method captures host-level communication behaviors by leveraging multiple entropies, focusing on characteristic patterns of IoT devices, such as periodic communication with specific servers during normal operation. This enables the reduction of computational resources during detection processing while maintaining detection accuracy, even when using fewer features and lightweight machine learning algorithms. The evaluation results demonstrate that the proposed method achieves a maximum reduction of 99.7% (2916 milliseconds) in processing time and 86.4% (633 MiB) in memory usage while maintaining an intrusion detection accuracy of 99.97%, proving its feasibility in constrained environments comparable to IoT gateways.journal articl
A Systematic Review on Blockchain-Enabled eKYC: Leveraging SSI and DID for Secure and Efficient Identity Verification
No full textThe rapid evolution of digital identity verification demands solutions that balance security, privacy, and efficiency. The electronic know your customer (eKYC) is a technological integration for client identification. It automates the process, reducing costs related to traditional know your customer (KYC). This includes eliminating paper-based document management, reducing manpower needs, and minimizing human errors. This systematic literature review (SLR) uses the preferred reporting items for systematic reviews and meta-analyses (PRISMA) model to investigate the revolutionary potential of blockchain-based electronic KYC (eKYC), focusing on self-sovereign identity (SSI) and Decentralized Identifiers (DID). The evaluation summarizes the current state by critically assessing 44 selected research works from an initial pool of 367. Our findings show that decentralized eKYC improves security with tamper-proof credentials and cryptographic verification. SSI and DID give users control over their data and selective disclosure. However, there are key limitations: 1) a focus on financial applications, ignoring Internet of Things (IoT) integration; 2) a lack of comprehensive technical analysis on scalability and interoperability; and 3) limited real-world case studies on regulatory compliance and challenges. This work combines insights from research and industry, highlighting the need for regulatory collaboration, hybrid architectures for scalability, and user-centric design. In addition, most identity management solutions are based on Ethereum (33%), followed by Hyperledger (18%). Around 51% of solutions use smart contracts, with banking (23%) and the financial industries (19%) being the primary adopters. It emphasizes the importance of standardized eKYC protocols, technical evaluations, and interdisciplinary collaboration for practical adoption across sectors.journal articl
The transcriptional repressors IAA5 and IAA29 participate in DNA damage-induced stem cell death in Arabidopsis roots
No full textPlants generate organs continuously during postembryonic development. Thus, their ability to preserve stem cells in changing environments is crucial for their survival. Genotoxic stress threatens genome stability in all somatic cells. However, in the meristem, only the stem cells actively die in response to DNA damage, followed by stem cell replenishment that guarantees genome stability in these cells. Cytokinin biosynthesis-induced inhibition of downward auxin flow participates in DNA damage-induced stem cell death in roots. Without this system, stem cell death occurs at a reduced but significant level, suggesting another mechanism governing the DNA damage response in stem cells. Here, we demonstrate that in response to DNA double-strand breaks, the AUXIN/INDOLE-3-ACETIC ACID (Aux/IAA) family members IAA5 and IAA29, encoding negative regulators of auxin signaling, are induced in Arabidopsis (Arabidopsis thaliana) roots. The transcription factor SUPPRESSOR OF GAMMA RESPONSE 1 directly induces their expression as an active response to DNA damage. In the iaa5 iaa29 double mutant, DNA damage-induced stem cell death is greatly suppressed, while it is fully restored by the expression of a stable form of IAA5 in vascular stem cells. Our genetic data reveal that reduced auxin signaling around the stem cell niche, caused by IAA5 and IAA29 induction and enhanced cytokinin biosynthesis, is a prerequisite for cell death induction, thus playing a central role in maintaining genome integrity in root stem cells.journal articl