The Effect Of Data Augmentation-Induced Class-Specific Bias Is Influenced By Data, Regularization | HackerNoon
Briefly

from Hackernoon6 months ago
The practical framework established for replicating experiments assesses the impact of data augmentation on model performance and class-specific bias, highlighting the need for scrutiny in design.
Hackernoonhttps://hackernoon.com/the-effect-of-data-augmentation-induced-class-specific-bias-is-influenced-by-data-regularization
Our data-centric analysis revealed that data augmentation strategies, particularly Random Cropping and Random Horizontal Flip, can induce varying levels of class-specific bias across different datasets.
Hackernoonhttps://hackernoon.com/the-effect-of-data-augmentation-induced-class-specific-bias-is-influenced-by-data-regularization
Read at Hackernoon
#data-augmentation #computer-vision #class-specific-bias #model-performance #machine-learning
[
|
]