Skip to content

Latest commit

 

History

History
36 lines (25 loc) · 1.83 KB

intrinsic_issues.md

File metadata and controls

36 lines (25 loc) · 1.83 KB

Intrinsic Issues of AL

Current intrinsic issues in the AL system and the solutions. These issues are not from the changes in assumptions but from the nature of AL.

AL Bias

The selection of AL introduces bias to the system.

Analysis

  • On Statistical Bias In Active Learning: How and When to Fix It [2021, ICLR]: Active learning can be helpful not only as a mechanism to reduce variance as it was originally designed, but also because it introduces a bias that can be actively helpful by regularizing the model.
  • Addressing Bias in Active Learning with Depth Uncertainty Networks... or Not [2021, Arxiv]: Reducing "AL bias" doesn't bring improvement on the low "overfitting bias" model DUN. When eliminate the "AL bias" with importance weights, we always pay the price of additional variance ("overfitting bias").

Adaptive loss

Thus, several models try to solve the overfitting at the beginning to improve the overall performance. It could be done by designing an adaptive loss.

Resampling

Resample the selected uncertainty data based on feature matching to alleviate the problem of data bias.

Resample:

  • Unsupervised Fusion Feature Matching for Data Bias in Uncertainty Active Learning [2022, TNNLS]

Out-of-Distribution (OOD) Detection:

  • Energy-based Out-of-distribution Detection [2020, NeurIPS]
  • Active Incremental Learning for Health State Assessment of Dynamic Systems With Unknown Scenarios [2022, IEEE Transactions on Industrial Informatics]