AI2030 Responsible AI Tools Marketplace

Key Takeaways

GANs can amplify existing biases: The paper demonstrates that when machine learning models use synthetic data generated by GANs (a common AI technique), these systems tend to amplify biases already present in the original training data, potentially making discrimination worse rather than better.
A practical solution for multiple AI systems: The researchers developed a fairness pipeline that works across different types of GAN architectures, addressing a key challenge in responsible AI—most previous fairness solutions only worked with specific AI systems, making them difficult to implement broadly across industries like banking and healthcare.
Removing "bias-inducing samples" improves fairness: The core innovation involves pre-processing training data to identify and remove samples that introduce bias, allowing GANs to focus on more representative and informative data. Testing on real datasets showed this approach generated fairer synthetic data while sometimes even improving overall model performance.
Addresses real-world AI challenges: The method tackles practical problems organizations face when deploying fair AI systems, including class imbalance (when some groups are underrepresented in data), limited training data, and privacy restrictions that prevent access to real sensitive information.
Fairness and performance aren't always trade-offs: Contrary to assumptions that fairness improvements require sacrificing accuracy, the paper's experiments suggest that generating fairer synthetic data can sometimes improve both fairness and model performance simultaneously.

Abstract

With the rising adoption of Machine Learning across the domains like banking, pharmaceutical, ed-tech, etc, it has become utmost important to adopt responsible AI methods to ensure models are not unfairly discriminating against any group. Given the lack of clean training data, generative adversarial techniques are preferred to generate synthetic data with several state-of-the-art architectures readily available across various domains from unstructured data such as text, images to structured datasets modelling fraud detection and many more. These techniques overcome several challenges such as class imbalance, limited training data, restricted access to data due to privacy issues. Existing work focusing on generating fair data either works for a certain GAN architecture or is very difficult to tune across the GANs. In this paper, we propose a pipeline to generate fairer synthetic data independent of the GAN architecture. The proposed paper utilizes a pre-processing algorithm to identify and remove bias inducing samples. In particular, we claim that while generating synthetic data most GANs amplify bias present in the training data but by removing these bias inducing samples, GANs essentially focuses more on real informative samples. Our experimental evaluation on two open-source datasets demonstrates how the proposed pipeline is generating fair data along with improved performance in some cases.

Key Takeaways

GANs can amplify existing biases: The paper demonstrates that when machine learning models use synthetic data generated by GANs (a common AI technique), these systems tend to amplify biases already present in the original training data, potentially making discrimination worse rather than better.
A practical solution for multiple AI systems: The researchers developed a fairness pipeline that works across different types of GAN architectures, addressing a key challenge in responsible AI—most previous fairness solutions only worked with specific AI systems, making them difficult to implement broadly across industries like banking and healthcare.
Removing "bias-inducing samples" improves fairness: The core innovation involves pre-processing training data to identify and remove samples that introduce bias, allowing GANs to focus on more representative and informative data. Testing on real datasets showed this approach generated fairer synthetic data while sometimes even improving overall model performance.
Addresses real-world AI challenges: The method tackles practical problems organizations face when deploying fair AI systems, including class imbalance (when some groups are underrepresented in data), limited training data, and privacy restrictions that prevent access to real sensitive information.
Fairness and performance aren't always trade-offs: Contrary to assumptions that fairness improvements require sacrificing accuracy, the paper's experiments suggest that generating fairer synthetic data can sometimes improve both fairness and model performance simultaneously.

Abstract

FairGen: Fair Synthetic Data Generation

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links

FairGen: Fair Synthetic Data Generation

Key Takeaways

Abstract

Paper Info

Responsible AI Topics

Links