Differential Privacy, Linguistic Fairness, and Training Data Influence: Impossibility and Possibility Theorems for Multilingual Language Models

Abstract

Language models such as mBERT, XLM-R, and BLOOM aim to achieve multilingual generalization or compression to facilitate transfer to a large number of (potentially unseen) languages. However, these models should ideally also be private, linguistically fair, and transparent, by relating their predictions to training data. Can these requirements be simultaneously satisfied? We show that multilingual compression and linguistic fairness are compatible with differential privacy, but that differential privacy is at odds with training data influence sparsity, an objective for transparency. We further present a series of experiments on two common NLP tasks and evaluate multilingual compression and training data influence sparsity under different privacy guarantees, exploring these trade-offs in more detail. Our results suggest that we need to develop ways to jointly optimize for these objectives in order to find practical trade-offs.

ICML 2023

References in corpus (9)

• Similarity of Neural Network Representations Revisited
• Universal Dependencies v2: An Evergrowing Multilingual Treebank Collection
• Differentially Private Learning Needs Better Features (or Much More Data)
• Data Privacy and Trustworthy Machine Learning
• What does it mean to be language-agnostic? Probing multilingual sentence encoders for typological properties
• Input Similarity from the Neural Network Perspective
• Fairness risk measures
• Enhancing Cross-lingual Transfer by Manifold Mixup
• The impossibility of "fairness": a generalized impossibility result for decisions