We study the settings for which deep contextual embeddings (e.g., BERT) ...
Many industrial machine learning (ML) systems require frequent retrainin...
Compressing word embeddings is important for deploying NLP models in
mem...
We investigate how to train kernel approximation methods that generalize...
We study large-scale kernel methods for acoustic modeling in speech
reco...
We study large-scale kernel methods for acoustic modeling and compare to...
The computational complexity of kernel methods has often been a major ba...