13197 research outputs found
Sort by
Study on the Enhancement of Security, Privacy, and Scalability in IoT Systems with Practical Blockchain-Based Access Control
Get PDF奈良先端科学技術大学院大学博士（理学）doctoral thesi
オンライン シミン シャカイ ニ ムケテ: フホウコウイトシテノヒボウチュウショウノジドウケンシュツギジュツのジレイケンキュウ
Get PDF奈良先端科学技術大学院大学博士（工学）doctoral thesi
End-to-end Simultaneous Speech Translation with Style Tags using Human Simultaneous Interpretation Data
No full textSimultaneous speech translation (SimulST) translates speech incrementally, requiring a monotonic input-output correspondence to reduce latency. This is particularly challenging for distant language pairs, such as English and Japanese, as most SimulST models are trained using offline speech translation (ST) data, where the entire speech input is observed during translation. In simultaneous interpretation (SI), a simultaneous interpreter translates source language speech into target language speech without waiting for the speaker to finish speaking. Therefore, the SimulST model can learn SI-style translations using SI data. However, owing to the limited availability of SI data, fine-tuning an offline ST model using SI data may result in overfitting. To address this problem, we propose an efficient training method for the speech-to-text SimulST model using a combination of small SI and relatively large offline ST data. We trained a single model with mixed data by incorporating style tags to instruct the model to generate either SI or offline-style outputs. This approach, called mixed fine-tuning with style tags, can be extended further using the multistage self-training approach. In this case, we use the trained model to generate pseudo-SI data. Our experimental results for several test sets demonstrated that our models trained using mixed fine-tuning and multistage self-training outperformed baselines across various latency ranges.journal articl
Material Segmentation Using 1-D Convolutional Neural Network With Transient Histogram
Get PDFThis study introduces a method for material classification using transient histograms obtained via a single-photon avalanche diode (SPAD) sensor. Temporal resolution in optical sensing plays a crucial role in material classification and surface segmentation, particularly for distinguishing materials with similar visual properties. In this study, SPAD sensors were utilized to capture transient histograms with temporal resolutions ranging from 13 picoseconds to 208 picoseconds, enabling precise extraction of temporal signatures for various materials. A comparative evaluation of classification techniques, including one-dimensional convolutional neural networks (1-D CNN), random forest (RF), support vector classifier (SVC), and k-nearest neighbors (KNN), was conducted to assess the impact of temporal resolution and exposure time on classification accuracy. 1-D CNN achieved the highest classification accuracy of 99.25% at a temporal resolution of 13 ps and an exposure time of 0.09 s, significantly outperforming other methods. Additionally, the proposed SPAD-based system was evaluated for material segmentation on non-planar surfaces. In a real-world experiment, 1-D CNN achieved an overall accuracy of 87.5% in differentiating visually similar materials, demonstrating the effectiveness of transient histograms for material classification where conventional RGB-based methods fail. These findings highlight the potential of SPAD sensors combined with advanced classification techniques to enhance material classification and segmentation, providing a versatile framework for applications in robotics, computer vision, and optical sensing.journal articl
The Role of GitHub Biographies: Impact on Pull Request Acceptance and Recruitment Decisions
Get PDF奈良先端科学技術大学院大学修士（工学）master thesi
Dockerfile プリプロセッサ ガ Docker イメージ カイハツ ホシュ カツドウ エ アタエル エイキョウ ノ ブンセキ
Get PDF奈良先端科学技術大学院大学修士（工学）master thesi
Sensor Pose Estimation from Low-Resolution SPAD Sensor Measurements
Get PDF奈良先端科学技術大学院大学修士（工学）master thesi
Integrating Traditional Medicine and Modern Techniques to Discover Novel Antibiotics: A Machine Learning Approach
Get PDF奈良先端科学技術大学院大学博士（工学）doctoral thesi
