Differential privacy (DP) data synthesizers are increasingly proposed to afford public release of sensitive information, offering theoretical guarantees for privacy (and, in some cases, utility), but limited empirical evidence of utility in practical settings. Utility is typically measured as the error on representative proxy tasks, such as descriptive statistics, multivariate correlations, the accuracy of trained classifiers, or performance over a query workload. The ability for these results to generalize to practitioners' experience has been questioned in a number of settings, including the U.S. Census. In this paper, we propose an evaluation methodology for synthetic data that avoids assumptions about the representativeness of proxy tasks, instead measuring the likelihood that published conclusions would change had the authors used synthetic data, a condition we call epistemic parity. Our methodology consists of reproducing empirical conclusions of peer-reviewed papers on real, publicly available data, then re-running these experiments a second time on DP synthetic data and comparing the results. We instantiate our methodology over a benchmark of recent peer-reviewed papers that analyze public datasets in the ICPSR social science repository. We model quantitative claims computationally to automate the experimental workflow, and model qualitative claims by reproducing visualizations and comparing the results manually. We then generate DP synthetic datasets using multiple state-of-the-art mechanisms, and estimate the likelihood that these conclusions will hold. We find that, for reasonable privacy regimes, state-of-the-art DP synthesizers are able to achieve high epistemic parity for several papers in our benchmark. However, some papers, and particularly some specific findings, are difficult to reproduce for any of the synthesizers. Given these results, we advocate for a new class of mechanisms that can reorder the priorities for DP data synthesis: favor stronger guarantees for utility (as measured by epistemic parity) and offer privacy protection with a focus on application-specific threat models and risk-assessment.

We describe a customizable benchmark for hierarchical and ontological multi-label classification, a task where labels are equipped with a graph structure and data items can be assigned multiple labels. We find that current benchmarks do not adequately represent the problem space, casting doubt on the generalizability of current results. We consider three dimensions of the problem space: context (availability of rich features on the data and labels), distribution of labels over data, and graph structure. For context, the lack of complex features on the labels (and in some cases, the data) artificially prevent the use of modern representation learning techniques as an appropriate baseline. For distribution, we find the long tail of labels over data constitute a few-shot learning problem that artificially confounds the results: for most common benchmarks, over 40% of the labels have fewer than 5 data points in the training set. For structure, we find that the correlation between performance and the height of the tree can explain some of the variation in performance, informing practical utility. In this paper, we demonstrate how the lack of diversity in benchmarks can confound performance analysis, then present a declarative query system called Ontologue for generating custom benchmarks with specific properties, then use this system to design 4 new benchmarks extracted from DBPedia that better represent the problem space. We evaluate state-of-the-art algorithms on both existing and new benchmarks and show that the performance conclusions can vary significantly depending on the dimensions we consider. We intend the system and derived benchmarks to improve the analysis of generalizability for these problems.

This paper reports on the making of an interactive demo to illustrate algorithmic bias in facial recognition. Facial recognition technology has been demonstrated to be more likely to misidentify women and minoritized people. This risk, among others, has elevated facial recognition into policy discussions across the country, where many jurisdictions have already passed bans on its use. Whereas scholarship on the disparate impacts of algorithmic systems is growing, general public awareness of this set of problems is limited in part by the illegibility of machine learning systems to non-specialists. Inspired by discussions with community organizers advocating for tech fairness issues, we created the Face Mis-ID Demo to reveal the algorithmic functions behind facial recognition technology and to demonstrate its risks to policymakers and members of the community. In this paper, we share the design process behind this interactive demo, its form and function, and the design decisions that honed its accessibility, toward its use for improving legibility of algorithmic systems and awareness of the sources of their disparate impacts.