When constructing models that learn from noisy labels produced by multiple annotators, it is important to accurately estimate the reliability of annotators. Annotators may provide labels of inconsistent quality due to their varying expertise and reliability in a domain. Previous studies have mostly focused on estimating each annotator's overall reliability on the entire annotation task. However, in practice, the reliability of an annotator may depend on each specific instance. Only a limited number of studies have investigated modelling perinstance reliability and these only considered binary labels. In this paper, we propose an unsupervised model which can handle both binary and multi-class labels. It can automatically estimate the per-instance reliability of each annotator and the correct label for each instance. We specify our model as a probabilistic model which incorporates neural networks to model the dependency between latent variables and instances. For evaluation, the proposed method is applied to both synthetic and real data, including two labelling tasks: text classification and textual entailment. Experimental results demonstrate our novel method can not only accurately estimate the reliability of annotators across different instances, but also achieve superior performance in predicting the correct labels and detecting the least reliable annotators compared to stateof-the-art baselines. 1

Unsupervised methods tend to discover highly speaker-specific representations of speech. We propose a method for improving the quality of posteriorgrams generated from an unsupervised model through partitioning of the latent classes. We do this by training a sparse siamese model to find a linear transformation of the input posteriorgrams to lower-dimensional posteriorgrams. The siamese model makes use of same-category and differentcategory speech fragment pairs obtained by unsupervised term discovery. After training, the model is converted into an exact partitioning of the posteriorgrams. We evaluate the model on the minimal-pair ABX task in the context of the Zero Resource Speech Challenge. We are able to demonstrate that our method significantly reduces the dimensionality of standard Gaussian mixture model posteriorgrams, while still making them more robust to speaker variations. This suggests that the model may be viable as a general post-processing step to improve probabilistic acoustic features obtained by unsupervised learning.

No abstract