Previous work in fair machine learning has characterised the Fair Bayes Optimal Classifier (BOC) on a given distribution for both deterministic and randomized classifiers. We study the robustness of the Fair BOC to adversarial noise in the data distribution. Kearns & Li (1988) implies that the accuracy of the deterministic BOC without any fairness constraints is robust (Lipschitz) to malicious noise in the data distribution. We demonstrate that their robustness guarantee breaks down when we add fairness constraints. Hence, we consider the randomized Fair BOC, and our central result is that its accuracy is robust to malicious noise in the data distribution. Our robustness result applies to various fairness constraints---Demographic Parity, Equal Opportunity, Predictive Equality. Beyond robustness, we demonstrate that randomization leads to better accuracy and efficiency. We show that the randomized Fair BOC is nearly-deterministic, and gives randomized predictions on at most one data point, hence availing numerous benefits of randomness, while using very little of it.

We study person-level differentially private (DP) mean estimation in the case where each person holds multiple samples. DP here requires the usual notion of distributional stability when all of a person’s datapoints can be modified. Informally, if n people each have m samples from an unknown d-dimensional distribution with bounded k-th moments, we show thatpeople are necessary and sufficient to estimate the mean up to distance α in ℓ2-norm under ε-differential privacy (and its common relaxations). In the multivariate setting, we give computationally efficient algorithms under approximate DP and computationally inefficient algorithms under pure DP, and our nearly matching lower bounds hold for the most permissive case of approximate DP. Our computationally efficient estimators are based on the standard clip-and-noise framework, but the analysis for our setting requires both new algorithmic techniques and new analyses. In particular, our new bounds on the tails of sums of independent, vector-valued, bounded-moments random variables may be of interest.

Multiple Instance Regression (MIR) and Learning from Label Proportions (LLP) are useful learning frameworks, where the training data is partitioned into disjoint sets or bags, and only an aggregate label, i.e., bag-label for each bag is available to the learner. In the case of MIR, the bag-label is the label of an undisclosed instance from the bag, while in LLP, the bag-label is the mean of the bag’s labels. In this paper, we study for various loss functions in MIR and LLP, what is the optimal way to partition the dataset into bags such that the utility for downstream tasks like linear regression is maximized. We theoretically provide utility guarantees, and show that in each case, the optimal bagging strategy (approximately) reduces to finding an optimal clustering of the feature vectors and/or the labels with respect to natural objectives such as $k$-means. We also show that our bagging mechanisms can be made label-differentially private, incurring an additional utility error. We then generalize our results to the setting of Generalized Linear Models (GLMs). Finally, we experimentally validate our theoretical results.

A few years ago, it was shown that there exist basic statistical learning problems whose learnability can not be determined within ZFC [Ben-David, Hrubes, Moran, Shpilka, Yehudayoff, 2017]. Such independence, and the implied impossibility of characterizing learnability of a class by any combinatorial parameter, stems from the basic definitions viewing learners as arbitrary functions. That level of generality not only results in unprovability issues but is also problematic from the perspective of modeling practical machine learning, where learners and predictors are computable objects. In light of that, it is natural to consider learnability by algorithms that output computable predictors (both learners and predictors are then representable as finite objects). A recent study [Agarwal, Ananthakrishnan, Ben-David, Lechner and Urner, 2020] initiated a theory of such models of learning. It proposed the notion of CPAC learnability, by adding some basic computability requirements into a PAC learning framework. As a first step towards a characterization of learnability in the CPAC framework, Agarwal et al showed that CPAC learnability of a binary hypothesis class is not implied by the finiteness of its VC-dimension anymore, as far as proper learners are concerned. A major remaining open question is whether a similar result holds also for improper learning. Namely, does there exist a computable concept class consisting of computable classifiers, that has a finite VC-dimension but no computable learner can PAC learn it (even if the learner is not restricted to output a hypothesis that is a member of the class)? Another implied interesting question concerns coming up with combinatorial characterizations of learnability for computable learners.

As machine learning black boxes are increasingly being deployed in critical domains such as healthcare and criminal justice, there has been a growing emphasis on developing techniques for explaining these black boxes in a post hoc manner. In this work, we analyze two popular post hoc interpretation techniques: SmoothGrad which is a gradient based method, and a variant of LIME which is a perturbation based method. More specifically, we derive explicit closed form expressions for the explanations output by these two methods and show that they both converge to the same explanation in expectation, i.e., when the number of perturbed samples used by these methods is large. We then leverage this connection to establish other desirable properties, such as robustness, for these techniques. We also derive finite sample complexity bounds for the number of perturbations required for these methods to converge to their expected explanation. Finally, we empirically validate our theory using extensive experimentation on both synthetic and real-world datasets.

SARS-CoV-2 has infected over ∼165 million people worldwide causing Acute Respiratory Distress Syndrome (ARDS) and has killed ∼3.4 million people. Artificial Intelligence (AI) has shown to benefit in the biomedical image such as X-ray/Computed Tomography in diagnosis of ARDS, but there are limited AI-based systematic reviews (aiSR). The purpose of this study is to understand the Risk-of-Bias (RoB) in a non-randomized AI trial for handling ARDS using novel AtheroPoint-AI-Bias (AP(ai)Bias). Our hypothesis for acceptance of a study to be in low RoB must have a mean score of 80% in a study. Using the PRISMA model, 42 best AI studies were analyzed to understand the RoB. Using the AP(ai)Bias paradigm, the top 19 studies were then chosen using the raw-cutoff of 1.9. This was obtained using the intersection of the cumulative plot of "mean score vs. study" and score distribution. Finally, these studies were benchmarked against ROBINS-I and PROBAST paradigm. Our observation showed that AP(ai)Bias, ROBINS-I, and PROBAST had only 32%, 16%, and 26% studies, respectively in low-moderate RoB (cutoff>2.5), however none of them met the RoB hypothesis. Further, the aiSR analysis recommends six primary and six secondary recommendations for the non-randomized AI for ARDS. The primary recommendations for improvement in AI-based ARDS design inclusive of (i) comorbidity, (ii) inter-and intra-observer variability studies, (iii) large data size, (iv) clinical validation, (v) granularity of COVID-19 risk, and (vi) cross-modality scientific validation. The AI is an important component for diagnosis of ARDS and the recommendations must be followed to lower the RoB.

We initiate a study of learning with computable learners and computable output predictors. Recent results in statistical learning theory have shown that there are basic learning problems whose learnability can not be determined within ZFC. This motivates us to consider learnability by algorithms with computable output predictors (both learners and predictors are then representable as finite objects). We thus propose the notion of CPAC learnability, by adding some basic computability requirements into a PAC learning framework. As a first step towards a characterization, we show that in this framework learnability of a binary hypothesis class is not implied by finiteness of its VC-dimension anymore. We also present some situations where we are guaranteed to have a computable learner.