Yu-Wen Chen

Yu-Wen Chen

Auto-generated profile, based on publicly available information

Yu-Wen Chen

Research Center for Information Technology Innovation, Academia Sinica, Tainan, Taiwan

Research Focus & Metrics

Academic Career Length

5Years (2022-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
22 February 2022 in IEEE Access
Verified
Journal Article
Open access
YCYu-Wen Chen
Yu-Wen Chen
KHKuo-Hsuan Hung
Kuo-Hsuan Hung
YLYou-Jin Li
You-Jin Li
AKAlexander Chao-Fu Kang
Alexander Chao-Fu Kang
YLYa-Hsin Lai
Ya-Hsin Lai
YTYu Tsao
Yu Tsao
This study presents a deep learning-based speech signal-processing mobile application known as CITISEN. The CITISEN can perform three functions: speech enhancement (SE), model adaptation (MA), and background noise conversion (BNC), which allow CITISEN to be used as a platform for utilizing and evaluating SE models and flexibly extend the models to address various noise environments and users. For SE, CITISEN downloads pretrained SE models on the cloud server and then uses these models to effectively reduce noise components from prerecordings or instant recordings provided by users. When it encounters noisy speech signals with unknown speakers or noise types, the MA function allows CITISEN to improve the SE performance effectively. A few audio files of unseen speakers or noise types are recorded and uploaded to the cloud server and then used to adapt the pretrained SE model. Finally, for BNC, CITISEN removes the original background noise using an SE model and then mixes the processed speech signal with new background noise. The novel BNC function can evaluate SE performance under specific conditions, cover people’s tracks, and provide entertainment. The experimental results confirmed the effectiveness of SE, MA, and BNC functions. Compared with the noisy speech signals, the enhanced speech signals by SE achieved about 6% and 33% of improvements, respectively, in terms of short-time objective intelligibility (STOI) and perceptual evaluation of speech quality (PESQ). With MA, the STOI and PESQ could be further improved by approximately 6% and 11%, respectively. Note that the SE model and MA method are not limited to the ones described in this study and can be replaced with any SE model and MA method. Finally, the BNC experiment results indicated that the speech signals of original and converted backgrounds have a close scene identification accuracy and similar embeddings in an acoustic scene classification model. Therefore, the proposed BNC can effectively convert the background noise of a speech signal and be a data augmentation method when clean speech signals are unavailable.

31 October 2022 in IEEE Transactions on Emerging Topics in Computational Intelligence
Verified
Journal Article
Closed access
HHHsun-Ping Hsieh
Hsun-Ping Hsieh
BLBao-Hong Lai
Bao-Hong Lai
YCYu-Wen Chen
Yu-Wen Chen
YCYu-En Chang
Yu-En Chang
Selecting the optimal location for a new bank branch is challenging but worth studying due to its importance to growing a successful business. Bankers have been devoting a lot of effort to this issue. In recent years, the proliferation of multisource data in smart cities has called for more need for data-driven optimal location selection. Previous studies usually focus on mining different features and processing these data through domain expert knowledge. However, they fail to discover potential complicated factors interactions. Besides, most of these studies only use a single indicator to evaluate the candidate locations, which overlooks a lot of information, resulting in an inaccurate evaluation. In this study, we propose MATE, a Multi-task Attentive Tree-Enhanced model, to simultaneously predict multiple performance indicators of a bank branch's location. Our model takes advantage of the tree-based model, which can effectively extract cross features and provide interpretability according to inferred decision rules. In addition, we designed the financial embedding module and attentive interaction blocks, allowing MATE to obtain more complex and diverse features. Finally, extensive experiments on real-world bank datasets demonstrate the effectiveness of our method, which outperforms baseline and state-of-art methods from 17% to 41%.