IJCNN/WCCI 2020 Special Session on
Neural Network-based Uncertainty Quantification
Scope and Aim
The field of artificial intelligence-based uncertainty quantification has gained an overwhelming attention among researchers in recent years resulting in an arsenal of different methods. Researchers have reported its key benefits in critical applications such as autonomous vehicles, medical diagnostics, bioinformatics, safety systems, and energy networks. Despite recent progresses, there are several key shortcomings with the existing frameworks for neural network-based uncertainty quantification. Deep neural networks as practiced conventionally fully or partially ignore uncertainties associated with predictions. They overconfidently generate point predictions for rare or unusual cases without providing a measure of their confidence in those unreliable predictions. Deep learning shortcomings in reliable uncertainty quantification could lead to catastrophes for autonomous vehicles, initiate the wrong diagnosis and treatment potentially resulting in severe medical complications, or cause prolonged and widespread blackouts when integrating renewable energies into the energy grid. The massive impact of ignoring or improperly treating neural network predictive uncertainty estimates has raised profound questions as to how artificial intelligence models can be trusted in critical applications. Establishing and maintaining this trust is a mind-boggling challenge for neural network models.
The goal is to provide an in-depth discussion of the latest academic and industrial research findings of artificial intelligence-based uncertainty quantification. The session will get together prominent and upcoming scientist from around the world who are conducting research on neural network-based uncertainty quantification using various approaches.
Topics of interest include, but are not limited to:
- Uncertainty quantification for different neural network types (feedforward, CNN, recurrent, LSTM, encoders, etc)
- Bayesian methods for uncertainty quantification
- Ensemble-based uncertainty quantification
- Direct methods for uncertainty quantification
- Temperature scaling and calibration methods
- Variational inference methods for uncertainty quantification
- Uncertainty-aware decision making (classification and regression)
- 15 Jan 2020 Paper Submission Deadline
- 15 Mar 2020 Paper Acceptance Notification Date
- 15 April 2020 Final Paper Submission and Early Registration Deadline
- 19-24 July 2020 IEEE WCCI 2020, Glasgow, Scotland, UK
This special session will be held in 2020 International Joint Conference on Neural Networks-IJCNN (wcci2020.org/ijcnn-sessions/), a part of 2020 IEEE World Congress on Computational Intelligence (https://wcci2020.org/ ) (Glasgow, Scotland, United Kingdom, July 19-24, 2020).
All papers should be prepared according to the IJCNN 2020 policy and should be submitted electronically using the conference website (https://wcci2020.org/ ) .
To submit your paper to this special session, you will choose our special session on the submission page "Neural Network-based Uncertainty Quantification". All papers accepted and presented at IEEE IJCNN/WCCI 2020 will be included in the conference proceedings published by IEEE Explore, which are typically indexed by EI.
Prof Saeid Nahavandi, Deakin University, saeid.nahavandi [at] deakin.edu.au
A/Prof Abbas Khosravi, Deakin University, abbas.khosravi [at] deakin.edu.au
Prof Dipti Srinivasan, National University of Singapore, dipti [at] nus.edu.sg