Can Private Machine Learning Be Fair?

Authors: Joseph Rance, Filip Svoboda

AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We show that current SOTA methods for privately and fairly training models are unreliable in many practical scenarios. Specifically, we (1) introduce a new type of adversarial attack that seeks to introduce unfairness into private model training, and (2) demonstrate that the use of methods for training on private data that are robust to adversarial attacks often leads to unfair models, regardless of the use of fairness-enhancing training methods. ... Experimental results. We test the fairness attack for the datasets described in table 2 ... All experiments were performed on 2 NVIDIA RTX 2080 GPUs. We record the change in fairness after the attack is introduced for each dataset-defence combination. The attack is effective at introducing unfairness into all three tasks.
Researcher Affiliation Academia Joseph Rance, Filip Svoboda Department of Computer Science & Technology University of Cambridge Cambridge, United Kingdom EMAIL
Pseudocode No The paper describes algorithms and attacks in prose and mathematical notation (e.g., Theorem 1, equations for Fed Avg), but it does not present any clearly labeled pseudocode or algorithm blocks.
Open Source Code Yes Code https://github.com/Joseph-Rance/unfair-fl
Open Datasets Yes Experimental results. We test the fairness attack for the datasets described in table 2 (Becker and Kohavi 1996; Krizhevsky 2009; Pushshift 2017). These datasets were selected to cover a range of tasks and to provide clear comparison with previous work (Bagdasaryan et al. 2019; Bhagoji et al. 2019; Wang et al. 2020; Nguyen et al. 2023; Mc Mahan et al. 2023). ... Table 2: UCI Census, CIFAR-10, Reddit
Dataset Splits No The paper discusses data distribution among clients (e.g., 'i.i.d. data', 'log-normal label distribution across the clients', splitting inputs for groups), but it does not provide specific train/validation/test dataset splits (e.g., percentages, sample counts, or references to standard splits for these datasets) that are generally used for model evaluation.
Hardware Specification Yes All experiments were performed on 2 NVIDIA RTX 2080 GPUs.
Software Dependencies No The paper mentions models like ResNet-50 and LSTM and tokenizers like albert-base-v2, but it does not list specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes Table 2: We train clients for 10, 2, and 5 epochs on i.i.d. data, for a total of 40, 120, and 100 rounds for the Census, CIFAR, and Reddit datasets respectively. ... We select hyperparameters by performing a grid search over all reasonable combinations at multiple levels of granularity and present the median result across three trials in table 1.