Unsupervised metric learning aims to learn the discriminative representations by grouping similar examples in the absence of labels. Many unsupervised metric learning algorithms combine clustering-based pseudo-label generation with embedding fine-tuning. However, pseudo-labels can be unreliable and noisy. This could affect metric learning and degrade the quality of the learned representations. In this work, we propose an approach to reduce the negative effect of label noise on learning discriminative embeddings by using context and prediction uncertainty. In particular, we refine the pseudo-labels by aggregating information from neighbors. We propose a function to weigh the pairs, leveraging their prediction confidence and uncertainty. We modify the metric learning loss function to incorporate this weight. Experimental results demonstrate the effectiveness of our proposed method on standard datasets for metric learning.

Acoustic scenes (or soundscapes) can be composed of various background and foreground sound events. Classification of soundscapes have to deal with the overlap of these sound events. In this work, we propose to reduce this overlap by considering foreground sound events and background sound events as multiple views of the soundscape. Robust principal components analysis is used to decompose a soundscape into the background view and the foreground view. We can control the amount of information retained in this decomposition by using the subspace projection technique of nuisance attribute projection. We represent the audio samples as features from a convolutional neural network, and view-invariant representations are derived using a deep neural network based multi-view learning algorithm. Experimental results demonstrate the effectiveness of our proposed method on standard datasets for acoustic scene classification.