EM-Network: Oracle Guided Self-distillation for Sequence Learning

Ji Won Yoon, Sunghwan Ahn, Hyeonseung Lee, Minchan Kim, Seok Min Kim, Nam Soo Kim
Proceedings of the 40th International Conference on Machine Learning, PMLR 202:40111-40128, 2023.

Abstract

We introduce EM-Network, a novel self-distillation approach that effectively leverages target information for supervised sequence-to-sequence (seq2seq) learning. In contrast to conventional methods, it is trained with oracle guidance, which is derived from the target sequence. Since the oracle guidance compactly represents the target-side context that can assist the sequence model in solving the task, the EM-Network achieves a better prediction compared to using only the source input. To allow the sequence model to inherit the promising capability of the EM-Network, we propose a new self-distillation strategy, where the original sequence model can benefit from the knowledge of the EM-Network in a one-stage manner. We conduct comprehensive experiments on two types of seq2seq models: connectionist temporal classification (CTC) for speech recognition and attention-based encoder-decoder (AED) for machine translation. Experimental results demonstrate that the EM-Network significantly advances the current state-of-the-art approaches, improving over the best prior work on speech recognition and establishing state-of-the-art performance on WMT’14 and IWSLT’14.

Cite this Paper

BibTeX
@InProceedings{pmlr-v202-yoon23a, title = {{EM}-Network: Oracle Guided Self-distillation for Sequence Learning}, author = {Yoon, Ji Won and Ahn, Sunghwan and Lee, Hyeonseung and Kim, Minchan and Kim, Seok Min and Kim, Nam Soo}, booktitle = {Proceedings of the 40th International Conference on Machine Learning}, pages = {40111--40128}, year = {2023}, editor = {Krause, Andreas and Brunskill, Emma and Cho, Kyunghyun and Engelhardt, Barbara and Sabato, Sivan and Scarlett, Jonathan}, volume = {202}, series = {Proceedings of Machine Learning Research}, month = {23--29 Jul}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v202/yoon23a/yoon23a.pdf}, url = {https://proceedings.mlr.press/v202/yoon23a.html}, abstract = {We introduce EM-Network, a novel self-distillation approach that effectively leverages target information for supervised sequence-to-sequence (seq2seq) learning. In contrast to conventional methods, it is trained with oracle guidance, which is derived from the target sequence. Since the oracle guidance compactly represents the target-side context that can assist the sequence model in solving the task, the EM-Network achieves a better prediction compared to using only the source input. To allow the sequence model to inherit the promising capability of the EM-Network, we propose a new self-distillation strategy, where the original sequence model can benefit from the knowledge of the EM-Network in a one-stage manner. We conduct comprehensive experiments on two types of seq2seq models: connectionist temporal classification (CTC) for speech recognition and attention-based encoder-decoder (AED) for machine translation. Experimental results demonstrate that the EM-Network significantly advances the current state-of-the-art approaches, improving over the best prior work on speech recognition and establishing state-of-the-art performance on WMT’14 and IWSLT’14.} }
Endnote
%0 Conference Paper %T EM-Network: Oracle Guided Self-distillation for Sequence Learning %A Ji Won Yoon %A Sunghwan Ahn %A Hyeonseung Lee %A Minchan Kim %A Seok Min Kim %A Nam Soo Kim %B Proceedings of the 40th International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2023 %E Andreas Krause %E Emma Brunskill %E Kyunghyun Cho %E Barbara Engelhardt %E Sivan Sabato %E Jonathan Scarlett %F pmlr-v202-yoon23a %I PMLR %P 40111--40128 %U https://proceedings.mlr.press/v202/yoon23a.html %V 202 %X We introduce EM-Network, a novel self-distillation approach that effectively leverages target information for supervised sequence-to-sequence (seq2seq) learning. In contrast to conventional methods, it is trained with oracle guidance, which is derived from the target sequence. Since the oracle guidance compactly represents the target-side context that can assist the sequence model in solving the task, the EM-Network achieves a better prediction compared to using only the source input. To allow the sequence model to inherit the promising capability of the EM-Network, we propose a new self-distillation strategy, where the original sequence model can benefit from the knowledge of the EM-Network in a one-stage manner. We conduct comprehensive experiments on two types of seq2seq models: connectionist temporal classification (CTC) for speech recognition and attention-based encoder-decoder (AED) for machine translation. Experimental results demonstrate that the EM-Network significantly advances the current state-of-the-art approaches, improving over the best prior work on speech recognition and establishing state-of-the-art performance on WMT’14 and IWSLT’14.
APA
Yoon, J.W., Ahn, S., Lee, H., Kim, M., Kim, S.M. & Kim, N.S.. (2023). EM-Network: Oracle Guided Self-distillation for Sequence Learning. Proceedings of the 40th International Conference on Machine Learning, in Proceedings of Machine Learning Research 202:40111-40128 Available from https://proceedings.mlr.press/v202/yoon23a.html.

Related Material