Learning Debiased Representation via Disentangled Feature Augmentation

07/03/2021
by   Eungyeup Kim, et al.
0

Image classification models tend to make decisions based on peripheral attributes of data items that have strong correlation with a target variable (i.e., dataset bias). These biased models suffer from the poor generalization capability when evaluated on unbiased datasets. Existing approaches for debiasing often identify and emphasize those samples with no such correlation (i.e., bias-conflicting) without defining the bias type in advance. However, such bias-conflicting samples are significantly scarce in biased datasets, limiting the debiasing capability of these approaches. This paper first presents an empirical analysis revealing that training with "diverse" bias-conflicting samples beyond a given training set is crucial for debiasing as well as the generalization capability. Based on this observation, we propose a novel feature-level data augmentation technique in order to synthesize diverse bias-conflicting samples. To this end, our method learns the disentangled representation of (1) the intrinsic attributes (i.e., those inherently defining a certain class) and (2) bias attributes (i.e., peripheral attributes causing the bias), from a large number of bias-aligned samples, the bias attributes of which have strong correlation with the target variable. Using the disentangled representation, we synthesize bias-conflicting samples that contain the diverse intrinsic attributes of bias-aligned samples by swapping their latent features. By utilizing these diversified bias-conflicting features during the training, our approach achieves superior classification accuracy and debiasing results against the existing baselines on both synthetic as well as real-world datasets.

READ FULL TEXT

page 6

page 9

page 12

page 16

research
05/29/2022

BiasEnsemble: Revisiting the Importance of Amplifying Bias for Debiasing

In image classification, "debiasing" aims to train a classifier to be le...
research
08/23/2021

BiaSwap: Removing dataset bias with bias-tailored swapping augmentation

Deep neural networks often make decisions based on the spurious correlat...
research
09/16/2022

Less is Better: Recovering Intended-Feature Subspace to Robustify NLU Models

Datasets with significant proportions of bias present threats for traini...
research
09/28/2022

Debiasing Graph Neural Networks via Learning Disentangled Causal Substructure

Most Graph Neural Networks (GNNs) predict the labels of unseen graphs by...
research
05/31/2022

Mitigating Dataset Bias by Using Per-sample Gradient

The performance of deep neural networks is strongly influenced by the tr...
research
03/24/2022

Intrinsic Bias Identification on Medical Image Datasets

Machine learning based medical image analysis highly depends on datasets...
research
07/06/2020

Learning from Failure: Training Debiased Classifier from Biased Classifier

Neural networks often learn to make predictions that overly rely on spur...

Please sign up or login with your details

Forgot password? Click here to reset