Adversarial Removal of Gender from Deep Image Representations

by   Tianlu Wang, et al.

In this work we analyze visual recognition tasks such as object and action recognition, and demonstrate the extent to which these tasks are correlated with features corresponding to a protected variable such as gender. We introduce the concept of natural leakage to measure the intrinsic reliance of a task on a protected variable. We further show that machine learning models of visual recognition trained for these tasks tend to exacerbate the reliance on gender features. To address this, we use adversarial training to remove unwanted features corresponding to protected variables from intermediate representations in a deep neural network. Experiments on two datasets: the COCO dataset (objects), and the imSitu dataset (actions), show reductions in the extent to which models rely on gender features while maintaining most of the accuracy of the original models. These results even surpass a strong baseline that blurs or removes people from images using ground-truth annotations. Moreover, we provide convincing interpretable visual evidence through an autoencoder-augmented model showing that this approach is performing semantically meaningful removal of gender features, and thus can also be used to remove gender attributes directly from images.


page 1

page 2

page 8


Training privacy-preserving video analytics pipelines by suppressing features that reveal information about private attributes

Deep neural networks are increasingly deployed for scene analytics, incl...

Right for the Right Reason: Training Agnostic Networks

We consider the problem of a neural network being requested to classify ...

Reducing Overlearning through Disentangled Representations by Suppressing Unknown Tasks

Existing deep learning approaches for learning visual features tend to o...

Gender Artifacts in Visual Datasets

Gender biases are known to exist within large-scale visual datasets and ...

Debiasing Convolutional Neural Networks via Meta Orthogonalization

While deep learning models often achieve strong task performance, their ...

Adversarial Concept Erasure in Kernel Space

The representation space of neural models for textual data emerges in an...

Please sign up or login with your details

Forgot password? Click here to reset