Paper: | SLP-P5.7 |
Session: | End-to-end Speech Recognition III: Source Integration and Knowledge Transfer |
Session Time: | Wednesday, May 15, 08:30 - 10:30 |
Presentation Time: | Wednesday, May 15, 08:30 - 10:30 |
Presentation: |
Poster
|
Topic: |
Speech Processing: Large Vocabulary Continuous Recognition/Search |
Paper Title: |
KNOWLEDGE DISTILLATION USING OUTPUT ERRORS FOR SELF-ATTENTION END-TO-END MODELS |
Authors: |
Ho-Gyeong Kim; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Hwidong Na; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Hoshik Lee; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Jihyun Lee; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Tae Gyoon Kang; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Min-Joong Lee; Samsung Advanced Institute of Technology, Samsung Electronics | | |
| Young Sang Choi; Samsung Advanced Institute of Technology, Samsung Electronics | | |