Computer Science > Neural and Evolutionary Computing
[Submitted on 27 Nov 2018 (this version), latest version 20 Sep 2019 (v3)]
Title:Uncertainty aware multimodal activity recognition with Bayesian inference
View PDFAbstract:Deep neural networks (DNNs) provide state-of-the-art results for a multitude of applications, but the use of DNNs for multimodal audiovisual applications is still an unsolved problem. The current approaches that combine audiovisual information do not consider inherent uncertainty or leverage true classification confidence associated with each modality in the final decision. Our contribution in this work is to apply Bayesian variational inference to DNNs for audiovisual activity recognition and quantify model uncertainty along with principled confidence. We propose a novel approach that combines deterministic and variational layers to estimate model uncertainty and principled confidence. Our experiments with in- and out-of-distribution samples selected from a subset of the Moments-in-Time (MiT) dataset show more reliable confidence measure as compared to the non-Bayesian baseline. We also demonstrate the uncertainty estimates obtained from this framework can identify out-of-distribution data on the UCF101 and MiT datasets. In the multimodal setting, the proposed framework improved precision-recall AUC by 14.4% on the subset of MiT dataset as compared to non-Bayesian baseline.
Submission history
From: Mahesh Subedar [view email][v1] Tue, 27 Nov 2018 04:51:54 UTC (1,305 KB)
[v2] Mon, 10 Jun 2019 06:01:04 UTC (1,747 KB)
[v3] Fri, 20 Sep 2019 05:35:30 UTC (1,768 KB)
Bibliographic and Citation Tools
Bibliographic Explorer (What is the Explorer?)
Litmaps (What is Litmaps?)
scite Smart Citations (What are Smart Citations?)
Code, Data and Media Associated with this Article
alphaXiv (What is alphaXiv?)
CatalyzeX Code Finder for Papers (What is CatalyzeX?)
DagsHub (What is DagsHub?)
Gotit.pub (What is GotitPub?)
Hugging Face (What is Huggingface?)
Papers with Code (What is Papers with Code?)
ScienceCast (What is ScienceCast?)
Demos
Recommenders and Search Tools
Influence Flower (What are Influence Flowers?)
Connected Papers (What is Connected Papers?)
CORE Recommender (What is CORE?)
arXivLabs: experimental projects with community collaborators
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.