Sampling strategies in Siamese Networks for unsupervised speech representation learning
This work addresses the often-overlooked sampling component in Siamese networks for unsupervised speech representation learning, offering incremental but practical improvements for researchers in speech processing.
The paper systematically investigates how different sampling strategies for selecting same vs. different word pairs affect the performance of Siamese networks in unsupervised speech representation learning, showing that strategies incorporating Zipf's Law, speaker distribution, and pair proportions significantly improve results, particularly with word frequency compression across training variations, and applying this to unsupervised algorithms yields state-of-the-art improvements.
Recent studies have investigated siamese network architectures for learning invariant speech representations using same-different side information at the word level. Here we investigate systematically an often ignored component of siamese networks: the sampling procedure (how pairs of same vs. different tokens are selected). We show that sampling strategies taking into account Zipf's Law, the distribution of speakers and the proportions of same and different pairs of words significantly impact the performance of the network. In particular, we show that word frequency compression improves learning across a large range of variations in number of training pairs. This effect does not apply to the same extent to the fully unsupervised setting, where the pairs of same-different words are obtained by spoken term discovery. We apply these results to pairs of words discovered using an unsupervised algorithm and show an improvement on state-of-the-art in unsupervised representation learning using siamese networks.