L
Lujun Li
Researcher at Technische Universität München
Publications - 27
Citations - 129
Lujun Li is an academic researcher from Technische Universität München. The author has contributed to research in topics: Computer science & Hidden Markov model. The author has an hindex of 3, co-authored 15 publications receiving 23 citations.
Papers
More filters
Book ChapterDOI
CTC-Segmentation of Large Corpora for German End-to-End Speech Recognition
TL;DR: In this paper, an end-to-end Automatic Speech Recognition (ASR) system demonstrated the ability to outperform conventional hybrid DNN/HMM ASR models in terms of depth, parameters and model capacity.
Book ChapterDOI
CTC-Segmentation of Large Corpora for German End-to-end Speech Recognition
TL;DR: Recent end-to-end Automatic Speech Recognition (ASR) systems demonstrated the ability to outperform conventional hybrid DNN/HMM ASR, but these models also require more training data to achieve comparable performance.
Proceedings ArticleDOI
Teacher-free Distillation via Regularizing Intermediate Representation
TL;DR: TFD, a simple and effective Teacher-Free Distillation framework, which seeks to reuse the privileged features within the student network itself by squeezing feature knowledge in the deeper layers into the shallow ones by minimizing feature loss.
Book ChapterDOI
Self-Regulated Feature Learning via Teacher-free Feature Distillation
TL;DR: Li et al. as discussed by the authors proposed a teacher-free feature distillation framework, which reuses channel-wise and layer-wise meaningful features within the student to provide teacher-like knowledge without an additional model.
Proceedings ArticleDOI
NORM: Knowledge Distillation via N-to-One Representation Matching
TL;DR: NORMao et al. as mentioned in this paper proposed N-to-One Representation (NORM), which relies on a simple Feature Transform (FT) module consisting of two linear layers to preserve the intact information learnt by the teacher network.