Summary of Uncertainty Regularized Evidential Regression, by Kai Ye et al.
Uncertainty Regularized Evidential Regression
by Kai Ye, Tiejin Chen, Hua Wei, Liang Zhan
First submitted to arxiv on: 3 Jan 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 This paper presents an innovative approach called Evidential Regression Network (ERN) which combines deep learning with Dempster-Shafer’s theory for uncertainty quantification in regression tasks. The ERN framework employs specific activation functions to ensure non-negative values, but this constraint limits the model’s ability to learn from all training samples. The authors provide a theoretical analysis of this limitation and introduce an improvement to overcome it. They first identify the regions where models are ineffective at learning from the data, then analyze the ERN and its constraints in detail. Based on these insights, they propose a novel regularization term that enables the ERN to learn from the entire training set. Extensive experiments validate their theoretical findings and demonstrate the effectiveness of the proposed solution. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary This paper is about a new way to use computers to predict things and figure out how sure we are of our predictions. It’s called Evidential Regression Network, or ERN for short. The problem with ERNs is that they can’t learn from all the data because they have to be careful not to make negative predictions. This limits what they can do. The authors of this paper studied this limitation and came up with a way to fix it. They looked at where ERNs struggle, analyzed how they work, and then developed a new technique to help them learn from all the data. Their tests showed that their solution really works! |
Keywords
* Artificial intelligence * Deep learning * Regression * Regularization
