Meta-Learning for Speech Emotion Recognition Considering Ambiguity of Emotion Labels

Emotion labels in emotion recognition corpora are highly noisy and ambiguous, due to the annotators’ subjective perception of emotions. Such ambiguity may introduce errors in automatic classification and affect the overall performance. We therefore propose a dynamic label correction and sample contribution weight estimation model. Our model is based on a standard BLSTM model with attention with two extra parameters. The first learns a new corrected label distribution and aims to fix the inaccurate labels in the dataset. The other estimates the contribution of each sample to the training process and aims to ignore the ambiguous and noisy samples while giving higher weights to the clear ones. We train our model through an alternating optimization method, where in the first epoch we update the neural network parameters, and in the second we keep them fixed to update the label correction and sample importance parameters. When training and evaluating our model on the IEMOCAP dataset, we obtained a weighted accuracy (WA) and unweighted accuracy (UA) of 65.9% and 61.4%, respectively. This yielded an absolute improvement of 2.3% and 1.9%, respectively, compared to a BLSTM with attention baseline, trained on the corpus gold labels.