Training neural audio classifiers with few data

Code

• github.com/jordipons/neural-classifiers-with-few-audio
  OfficialIn papertf★ 0
• github.com/yagyapandeya/CNN-with-Few-Data-VGGish-
  tf★ 0

Abstract

We investigate supervised learning strategies that improve the training of neural network audio classifiers on small annotated collections. In particular, we study whether (i) a naive regularization of the solution space, (ii) prototypical networks, (iii) transfer learning, or (iv) their combination, can foster deep learning models to better leverage a small amount of training examples. To this end, we evaluate (i-iv) for the tasks of acoustic event recognition and acoustic scene classification, considering from 1 to 100 labeled examples per class. Results indicate that transfer learning is a powerful strategy in such scenarios, but prototypical networks show promising results when one does not count with external or validation data.

Tasks

Reproductions