HODGEPODGE: Sound Event Detection Based on Ensemble of Semi-Supervised Learning Methods

Abstract

In this paper, we present a method called HODGEPODGE 1 for largescale detection of sound events using weakly labeled, synthetic, and unlabeled data proposed in the Detection and Classification of Acoustic Scenes and Events (DCASE) 2019 challenge Task 4: Sound event detection in domestic environments. To perform this task, we adopted the convolutional recurrent neural networks (CRNN) as our backbone network. In order to deal with a small amount of tagged data and a large amounts of unlabeled in-domain data, we aim to focus primarily on how to apply semi-supervise learning methods efficiently to make full use of limited data. Three semi-supervised learning principles have been used in our system, including: 1) Consistency regularization applies data augmentation; 2) MixUp regularizer requiring that the predictions for a interpolation of two inputs is close to the interpolation of the prediction for each individual input; 3) MixUp regularization applies to interpolation between data augmentations. We also tried an ensemble of various models, which are trained by using different semisupervised learning principles. Our proposed approach significantly improved the performance of the baseline, achieving the event-based f-measure of 42.0% compared to 25.8% event-based f-measure of the baseline in the provided official evaluation dataset. Our submissions ranked third among 18 teams in the task 4. 1 HODGEPODGE has two layers of meanings. The first layer is the variety of training data involved in the method, including weakly labeled, synthetic, and unlabeled data. The second layer refers to several semi-supervised principles involved in our method.

PSLA: Improving Audio Tagging With Pretraining, Sampling, Labeling, and Aggregation

“…The goal of ensemble methods is to combine the predictions of several models to improve generalizability and robustness over any single model. Previously, ensemble of AEC models has been studied in in [14], [52], [53], [54], [15], [12], [4], but typically only one strategy is covered in each of these previous efforts. In this work, we use the simple voting algorithm, but compare The first strategy investigated is checkpoint averaging, whereby the output of checkpoint models at multiple epochs are averaged together.…”

Hear Me Out: Fusional Approaches for Audio Augmented Temporal Action Localization

“…Audio-based Localization: Speaker diarization [30,31] involves localization of speaker boundaries and grouping segments that belong to the same speaker. The DCASE Challenge examines sound event detection in domestic environments as one of the challenge tasks [32,33,34,35,36,37]. In our action localization setting, note that audio modality is unrestricted, i.e.…”

Sound Event Detection by Consistency Training and Pseudo-Labeling With Feature-Pyramid Convolutional Recurrent Neural Networks

“…We adopt the ideas of a prior work [15] which applied interpolation consistency training (ICT) to weakly-labeled semisupervised SED. ICT encourages the prediction at an interpolation of unlabeled points to be consistent with the interpolation of predictions at those points; that is,…”

“…It is intuitive that learning from interpolation samples can help the model discriminate samples that are ambiguous between two classes. Implementation of [15] replaces all input samples with interpolation samples and calculates the same loss function as the baseline. However, we find that original input samples can stabilize the model performance during training.…”