Transformer attention is reframed as kernel smoothing, and a product of symmetric kernels for features and positions achieves competitive performance on neural machine translation and sequence prediction.
Classical Structured Prediction Losses for Sequence to Sequence Learning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
There has been much recent work on training neural attention models at the sequence-level using either reinforcement learning-style methods or by optimizing the beam. In this paper, we survey a range of classical objective functions that have been widely used to train linear models for structured prediction and apply them to neural sequence to sequence models. Our experiments show that these losses can perform surprisingly well by slightly outperforming beam search optimization in a like for like setup. We also report new state of the art results on both IWSLT'14 German-English translation as well as Gigaword abstractive summarization. On the larger WMT'14 English-French translation task, sequence-level training achieves 41.5 BLEU which is on par with the state of the art.
fields
cs.LG 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Transformer Dissection: A Unified Understanding of Transformer's Attention via the Lens of Kernel
Transformer attention is reframed as kernel smoothing, and a product of symmetric kernels for features and positions achieves competitive performance on neural machine translation and sequence prediction.