Modeling and Simultaneously Removing Bias via Adversarial Neural Networks
This addresses bias issues in real-world systems like online advertising, offering a novel method for mitigating feedback-induced bias, though it is incremental as it builds on adversarial approaches.
The paper tackles bias in machine learning models caused by feedback loops in deployed systems, such as paid search auctions, by developing an Adversarial Neural Network (ANN) model that creates bias-invariant data representations, and demonstrates its success on synthetic and real-world data.
In real world systems, the predictions of deployed Machine Learned models affect the training data available to build subsequent models. This introduces a bias in the training data that needs to be addressed. Existing solutions to this problem attempt to resolve the problem by either casting this in the reinforcement learning framework or by quantifying the bias and re-weighting the loss functions. In this work, we develop a novel Adversarial Neural Network (ANN) model, an alternative approach which creates a representation of the data that is invariant to the bias. We take the Paid Search auction as our working example and ad display position features as the confounding features for this setting. We show the success of this approach empirically on both synthetic data as well as real world paid search auction data from a major search engine.