Summary of Deepuq: Assessing the Aleatoric Uncertainties From Two Deep Learning Methods, by Rebecca Nevin et al.
DeepUQ: Assessing the Aleatoric Uncertainties from two Deep Learning Methods
by Rebecca Nevin, Aleksandra Ćiprijanović, Brian D. Nord
First submitted to arxiv on: 13 Nov 2024
Categories
- Main: Machine Learning (cs.LG)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper compares two deep learning-based uncertainty quantification (UQ) techniques, Deep Ensembles (DE) and Deep Evidential Regression (DER), to assess the quality of aleatoric uncertainty estimates. The methods are evaluated on both zero-dimensional (0D) and two-dimensional (2D) data, exploring their performance for different data dimensionalities. The experiments involve injecting uncertainty on input and output variables, propagating uncertainty in input uncertain cases, and testing with three noise levels. The results show that the predicted aleatoric uncertainty scales with injected noise but is miscalibrated compared to true uncertainty for some DE and DER experiments. Specifically, post-facto calibration for these methods would be beneficial, particularly in high-noise and high-dimensional settings. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper compares two deep learning-based ways to measure uncertainty. These methods are tested on different types of data (simple or complex) to see how well they work. The results show that the methods can predict uncertainty, but they don’t always get it right. This is important because uncertainty matters in science and we need to understand it correctly. |
Keywords
* Artificial intelligence * Deep learning * Regression
