Our token-wise CNN with BERT features is ten times faster than RNNs in for shortening original sentences and preserving their key information. This can benefit, e.g., tools for language education. @anucecs @Data61news #MachineLearning #LanguageLearning h
225 followers
6,087 followers
Deep Residual Local Feature Learning for Speech Emotion Recognition https://t.co/3wSGhuGKz7
6,087 followers
Deep Residual Local Feature Learning for Speech Emotion Recognition https://t.co/3wSGhuGKz7
102 followers
RT @arxiv_org: Deep Residual Local Feature Learning for Speech Emotion Recognition. https://t.co/CaxjF4U21S https://t.co/tF0cS5gGKR