Large Scale Audiovisual Learning of Sounds with Weakly Labeled Data

Authors: Haytham M. Fayek, Anurag Kumar

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on the large scale sound events dataset, Audio Set, demonstrate the efficacy of the proposed model, which outperforms the single-modal models, and state-of-the-art fusion and multi-modal models. We achieve a mean Average Precision (m AP) of 46.16 on Audioset, outperforming prior state of the art by approximately +4.35 m AP (relative: 10.4%).
Researcher Affiliation Industry Facebook Reality Labs, Redmond, WA, USA {haythamfayek, anuragkr}@fb.com
Pseudocode No The paper describes model architectures and mathematical formulations but does not include any clearly labeled pseudocode or algorithm blocks.
Open Source Code No The paper does not contain any explicit statements about releasing source code or provide links to a code repository for the methodology described.
Open Datasets Yes Audioset [Gemmeke et al., 2017] is the largest dataset for sound events. The dataset provides You Tube videos for 527 sound events.
Dataset Splits Yes The training set comprises approximately 2 million videos, whereas the evaluation set comprises approximately 20, 000 videos. We sample approximately 25, 000 videos from the training set to use as the validation set.
Hardware Specification No The paper does not specify the exact hardware (e.g., GPU, CPU models, or cloud resources) used for running the experiments.
Software Dependencies No The paper mentions using the Adam optimizer but does not provide specific version numbers for any software dependencies or libraries.
Experiment Setup Yes The network in the video model is trained for 20 epochs using the Adam optimizer [Kingma and Ba, 2014] with a mini-batch size of 144. In the fusion experiments, all neural networks are trained using Adam for 100 epochs. The mini-batch size is set to 256. na, nv, and nav are all single layer networks with 512 units and sigmoid activations.