Fairness in Forecasting of Observations of Linear Dynamical Systems
arXiv:2209.05274 · doi:10.1613/jair.1.14050
Abstract
In machine learning, training data often capture the behaviour of multiple subgroups of some underlying human population. This behaviour can often be modelled as observations of an unknown dynamical system with an unobserved state. When the training data for the subgroups are not controlled carefully, however, under-representation bias arises. To counter under-representation bias, we introduce two natural notions of fairness in time-series forecasting problems: subgroup fairness and instantaneous fairness. These notions extend predictive parity to the learning of dynamical systems. We also show globally convergent methods for the fairness-constrained learning problems using hierarchies of convexifications of non-commutative polynomial optimisation problems. We also show that by exploiting sparsity in the convexifications, we can reduce the run time of our methods considerably. Our empirical results on a biased data set motivated by insurance applications and the well-known COMPAS data set demonstrate the efficacy of our methods.
arXiv commentJournal version of Zhou et al. [arXiv:2006.07315, AAAI 2021]
References in corpus (33)
- SMOTE: Synthetic Minority Over-sampling Technique
- Equality of Opportunity in Supervised Learning
- Counterfactual Fairness
- Inherent Trade-Offs in the Fair Determination of Risk Scores
- Bias and Discrimination in AI: a cross-disciplinary perspective
- AI Fairness 360: An Extensible Toolkit for Detecting, Understanding, and Mitigating Unwanted Algorithmic Bias
- Avoiding Discrimination through Causal Reasoning
- On Formalizing Fairness in Prediction with Machine Learning
- On the relation between accuracy and fairness in binary classification
- Chordal-TSSOS: a moment-SOS hierarchy that exploits term sparsity with chordal extension
- Near optimal finite time identification of arbitrary linear dynamical systems
- Learning Without Mixing: Towards A Sharp Analysis of Linear System Identification
- Robust Optimization for Fairness with Noisy Protected Groups
- Learning Linear Dynamical Systems with Semi-Parametric Least Squares
- Average Individual Fairness: Algorithms, Generalization and Experiments
- TSSOS: A Moment-SOS hierarchy that exploits term sparsity
- Algorithms for Fairness in Sequential Decision Making
- Spectral Filtering for General Linear Dynamical Systems
- Learning Fair Scoring Functions: Bipartite Ranking under ROC-based Fairness Constraints
- Recovering from Biased Data: Can Fairness Constraints Improve Accuracy?
- Assessing Fairness in the Presence of Missing Data
- Empirical observation of negligible fairness-accuracy trade-offs in machine learning for public policy
- Dynamic fairness - Breaking vicious cycles in automatic decision making
- Causal Conceptions of Fairness and their Consequences
- Representation Matters: Assessing the Importance of Subgroup Allocations in Training Data
- Fairness in Forecasting and Learning Linear Dynamical Systems
- Fairness without Imputation: A Decision Tree Approach for Fair Prediction with Missing Values
- Does enforcing fairness mitigate biases caused by subpopulation shift?
- Sample Complexity of Kalman Filtering for Unknown Systems
- Learning Linear Dynamical Systems via Spectral Filtering
- Exploiting term sparsity in Noncommutative Polynomial Optimization
- Online Learning of the Kalman Filter with Logarithmic Regret
- Protecting Consumers Against Personalized Pricing: A Stopping Time Approach