3 papers
cs.LG2021
Knowledge distillation from language model to acoustic model: a hierarchical multi-task learning approach
Mun-Hak Lee, Joon-Hyuk Chang
The remarkable performance of the pre-trained language model (LM) using self-supervised learning has led to a major paradigm shift in the study of natural language processing. In l…
cs.LG2021
Attribution Mask: Filtering Out Irrelevant Features By Recursively Focusing Attention on Inputs of DNNs
Jae-Hong Lee, Joon-Hyuk Chang
Attribution methods calculate attributions that visually explain the predictions of deep neural networks (DNNs) by highlighting important parts of the input features. In particular…
cs.SD2018
Augmenting Bottleneck Features of Deep Neural Network Employing Motor State for Speech Recognition at Humanoid Robots
Moa Lee, Joon Hyuk Chang
As for the humanoid robots, the internal noise, which is generated by motors, fans and mechanical components when the robot is moving or shaking its body, severely degrades the per…