Search

Your search keyword '"Lee DO"' showing total 6,758 results
Start Over Author "Lee DO"

Refine your results

more »

more »

more »

more »

more »

more »
6,758 results on '"Lee DO"'

1. Performance Improvement of Language-Queried Audio Source Separation Based on Caption Augmentation From Large Language Models for DCASE Challenge 2024 Task 9

Author

Lee, Do Hyun, Song, Yoonah, and Kim, Hong Kook

Subjects
Electrical Engineering and Systems Science - Audio and Speech Processing, Computer Science - Artificial Intelligence, Computer Science - Sound
Abstract

We present a prompt-engineering-based text-augmentation approach applied to a language-queried audio source separation (LASS) task. To enhance the performance of LASS, the proposed approach utilizes large language models (LLMs) to generate multiple captions corresponding to each sentence of the training dataset. To this end, we first perform experiments to identify the most effective prompts for caption augmentation with a smaller number of captions. A LASS model trained with these augmented captions demonstrates improved performance on the DCASE 2024 Task 9 validation set compared to that trained without augmentation. This study highlights the effectiveness of LLM-based caption augmentation in advancing language-queried audio source separation., Comment: DCASE 2024 Challenge Task 9, 4 pages

Published
2024

4. Atmospheric Humidity Underlies Irreproducibility of Formamidinium Lead Iodide Perovskites

Author

Park, Keonwoo, Tan, Shaun, Kodalle, Tim, Lee, Do‐Kyoung, Abdelsamie, Maged, Park, Ji‐Sang, Lee, Joo‐Hong, Jung, Sung‐Kwang, Ko, Jeong Hoon, Park, Nam‐Gyu, Sutter‐Fella, Carolin M, Yang, Yang, and Lee, Jin‐Wook

Subjects
Macromolecular and Materials Chemistry, Chemical Sciences, Physical Chemistry, crystallizations, halide perovskites, humidity, in situ studies, reproducibility, scalability, solar cells, stability, Physical Sciences, Engineering, Nanoscience & Nanotechnology, Chemical sciences, Physical sciences
Abstract

Metal halide perovskite solar cells (PSCs) are infamous for their batch-to-batch and lab-to-lab irreproducibility in terms of stability and performance. Reproducible fabrication of PSCs is a critical requirement for market viability and practical commercialization. PSC irreproducibility plagues all levels of the community; from institutional research laboratories, start-up companies, to large established corporations. In this work, the critical function of atmospheric humidity to regulate the crystallization and stabilization of formamidinium lead triiodide (FAPbI3) perovskites is unraveled. It is demonstrated that the humidity content during processing induces profound variations in perovskite stoichiometry, thermodynamic stability, and optoelectronic quality. Almost counterintuitively, it is shown that the presence of humidity is perhaps indispensable to reproduce phase-stable and efficient FAPbI3-based PSCs.

Published
2024

7. GIST-AiTeR Speaker Diarization System for VoxCeleb Speaker Recognition Challenge (VoxSRC) 2023

Author

Park, Dongkeon, Kim, Ji Won, Kim, Kang Ryeol, Lee, Do Hyun, and Kim, Hong Kook

Subjects
Electrical Engineering and Systems Science - Audio and Speech Processing
Abstract

This report describes the submission system by the GIST-AiTeR team for the VoxCeleb Speaker Recognition Challenge 2023 (VoxSRC-23) Track 4. Our submission system focuses on implementing diverse speaker diarization (SD) techniques, including ResNet293 and MFA-Conformer with different combinations of segment and hop length. Then, those models are combined into an ensemble model. The ResNet293 and MFA-Conformer models exhibited the diarization error rates (DERs) of 3.65% and 3.83% on VAL46, respectively. The submitted ensemble model provided a DER of 3.50% on VAL46, and consequently, it achieved a DER of 4.88% on the VoxSRC-23 test set., Comment: VoxSRC 2023 Track4

Published
2023

8. Effect of the Precursor Chemistry on the Crystallization of Triple Cation Mixed Halide Perovskites

Author

Singh, Mriganka, Abdelsamie, Maged, Li, Qihua, Kodalle, Tim, Lee, Do-Kyoung, Arnold, Simon, Ceratti, Davide R, Slack, Jonathan L, Schwartz, Craig P, Brabec, Christoph J, Tao, Shuxia, and Sutter-Fella, Carolin M

Subjects
Macromolecular and Materials Chemistry, Chemical Sciences, Physical Chemistry, Affordable and Clean Energy, Engineering, Materials, Chemical sciences
Abstract

Triple cation, mixed halide perovskite compositions have been reported to be more thermally stable, exhibit fewer phase impurities, and show higher power conversion efficiency and better reproducibility than single cation perovskites. In this work, we explain the formation of Cs0.05FA0.81MA0.14Pb(I0.85Br0.15)3 via a multimodal in situ study combining structural information from synchrotron grazing-incidence wide-angle X-ray scattering (GIWAXS) and optical properties from photoluminescence (PL) spectroscopy with density functional theory calculations (DFT). The focus here is on the effects of the solvent and antisolvent during crystallization. The predominantly used solvents N,N-dimethylformamide (DMF), dimethyl sulfoxide (DMSO), and the antisolvent chlorobenzene (CB) as well as the solvent-antisolvent-precursor interactions are investigated. Given the high elemental complexity and mutual interdependencies between solvent, antisolvent, and perovskite precursors, we found significant differences in the crystallization pathways. DMF-pure precursors show the formation of the DMF-containing intermediate phase and the nucleation of compositionally distinct perovskite phases, while when DMSO is added, only crystalline α- and δ-phases were found. In addition, the presence of DMSO helps the formation of α-perovskite. Coordination energy and bond order (BO) calculations support our experimental findings. Dripping of CB induces nucleation at room temperature, slows the α-phase formation rate, and appears to reduce the nucleation radius. These findings provide novel insights into solvent, antisolvent, and perovskite precursor interactions and their formation pathways. The complexity of interactions between solvents and reagents highlights the importance of understanding these effects to further improve the reproducibility and optimize processing conditions.

Published
2023

21. A pilot clinical study on the Traditional Korean Medicine treatment of Amyotrophic lateral sclerosis

Author

Kim Sung-chul, Na Won-min, Lim Na-ra, Lee Do-sang, Jang Eun-ha, and Song Bong-keun

Subjects
Amyotrophic lateral sclerosis(ALS), ALS Functional Rating Scale-Revised(ALSFRSR), Medical Research Council(MRC) Scale, Medicine, Miscellaneous systems and treatments, RZ409.7-999, Therapeutics. Pharmacology, RM1-950
Abstract

Objectives : This study was to investigate the effect of Oriental medical treatment on ALS. Methods : We investigated 12 ALS patients which were admitted to Gwang-Ju O.M. hospital from Oct. 14, 2008 to Nov. 14, 2008. All patients were treated by SAAM-acupuncture, herb medication, Bee venom Pharmacopuncture therapy, Needle-embedding therapy, etc. We evaluated patients using the Amyotrophic lateral sclerosis Functional Rating Scale-Revised(ALSFRS-R), Medical Research Council (MRC) Scale. Results : After 30 days, mean ALSFRS-R score of patients was improved from 28.42±7.83 to 29.08 ±7.99, and mean MRC Scale of patients was improved from 24.79±8.37 to 25.34±8.45. But in both cases, the variation was not statistically significant. After 30 days, mean ALSFRS-R score and mean MRC Scale of patients was more improved in subjects with bulbar-onset, onset age: 51-60yrs., disease duration: 24-48mo. And the results showed partially significant difference. Conclusions : We think that the results of this case be a pilot study that proves the effect of Oriental Medical treatment on ALS.

Published
2009
Full Text
View/download PDF

23. Context-based Virtual Adversarial Training for Text Classification with Noisy Labels

Author

Lee, Do-Myoung, Kim, Yeachan, and Seo, Chang-gyun

Subjects
Computer Science - Computation and Language, Computer Science - Artificial Intelligence, Computer Science - Machine Learning
Abstract

Deep neural networks (DNNs) have a high capacity to completely memorize noisy labels given sufficient training time, and its memorization, unfortunately, leads to performance degradation. Recently, virtual adversarial training (VAT) attracts attention as it could further improve the generalization of DNNs in semi-supervised learning. The driving force behind VAT is to prevent the models from overfitting data points by enforcing consistency between the inputs and the perturbed inputs. This strategy could be helpful in learning from noisy labels if it prevents neural models from learning noisy samples while encouraging the models to generalize clean samples. In this paper, we propose context-based virtual adversarial training (ConVAT) to prevent a text classifier from overfitting to noisy labels. Unlike the previous works, the proposed method performs the adversarial training at the context level rather than the inputs. It makes the classifier not only learn its label but also its contextual neighbors, which alleviates the learning from noisy labels by preserving contextual semantics on each data point. We conduct extensive experiments on four text classification datasets with two types of label noises. Comprehensive experimental results clearly show that the proposed method works quite well even with extremely noisy settings., Comment: 10 pages, Proceedings of Language Resources and Evaluation, 2022 (LREC'22)

Published
2022

Catalog

Books, media, physical & digital resources
See catalog results