Feedback Detection for Live Predictors

TL;DR

This paper introduces a causal inference-based method with local randomization to detect feedback loops in live predictors, demonstrated through a pilot study in a search engine system.

Contribution

It presents a novel approach to identify non-linear predictor feedback in real-world systems using local randomization techniques.

Findings

Effective detection of feedback loops demonstrated in a search engine system

Method can identify non-linear feedback in live predictors

Pilot study validates practical applicability

Abstract

A predictor that is deployed in a live production system may perturb the features it uses to make predictions. Such a feedback loop can occur, for example, when a model that predicts a certain type of behavior ends up causing the behavior it predicts, thus creating a self-fulfilling prophecy. In this paper we analyze predictor feedback detection as a causal inference problem, and introduce a local randomization scheme that can be used to detect non-linear feedback in real-world problems. We conduct a pilot study for our proposed methodology using a predictive system currently deployed as a part of a search engine.