Despite an increasing reliance on fully-automated algorithmic decision-making in our day-to-day lives, human beings still make highly consequential decisions. As frequently seen in business, healthcare, and public policy, recommendations produced by algorithms are provided to human decision-makers to guide their decisions. While there exists a fast-growing literature evaluating the bias and fairness of such algorithmic recommendations, an overlooked question is whether they help humans make better decisions. We develop a statistical methodology for experimentally evaluating the causal impacts of algorithmic recommendations on human decisions. We also show how to examine whether algorithmic recommendations improve the fairness of human decisions and derive the optimal decision rules under various settings. We apply the proposed methodology to preliminary data from the first-ever randomized controlled trial that evaluates the pretrial Public Safety Assessment (PSA) in the criminal justice system. A goal of the PSA is to help judges decide which arrested individuals should be released. On the basis of the preliminary data available, we find that providing the PSA to the judge has little overall impact on the judge's decisions and subsequent arrestee behavior. However, our analysis yields some potentially suggestive evidence that the PSA may help avoid unnecessarily harsh decisions for female arrestees regardless of their risk levels while it encourages the judge to make stricter decisions for male arrestees who are deemed to be risky. In terms of fairness, the PSA appears to increase the gender bias against males while having little effect on any existing racial differences in judges' decision. Finally, we find that the PSA's recommendations might be unnecessarily severe unless the cost of a new crime is sufficiently high.
相關內容
Just-in-time adaptive interventions (JITAIs) are time-varying adaptive interventions that use frequent opportunities for the intervention to be adapted--weekly, daily, or even many times a day. The micro-randomized trial (MRT) has emerged for use in informing the construction of JITAIs. MRTs can be used to address research questions about whether and under what circumstances JITAI components are effective, with the ultimate objective of developing effective and efficient JITAI. The purpose of this article is to clarify why, when, and how to use MRTs; to highlight elements that must be considered when designing and implementing an MRT; and to review primary and secondary analyses methods for MRTs. We briefly review key elements of JITAIs and discuss a variety of considerations that go into planning and designing an MRT. We provide a definition of causal excursion effects suitable for use in primary and secondary analyses of MRT data to inform JITAI development. We review the weighted and centered least-squares (WCLS) estimator which provides consistent causal excursion effect estimators from MRT data. We describe how the WCLS estimator along with associated test statistics can be obtained using standard statistical software such as R (R Core Team, 2019). Throughout we illustrate the MRT design and analyses using the HeartSteps MRT, for developing a JITAI to increase physical activity among sedentary individuals. We supplement the HeartSteps MRT with two other MRTs, SARA and BariFit, each of which highlights different research questions that can be addressed using the MRT and experimental design considerations that might arise.
In this work we consider the well-known Secretary Problem -- a number $n$ of elements, each having an adversarial value, are arriving one-by-one according to some random order, and the goal is to choose the highest value element. The decisions are made online and are irrevocable -- if the algorithm decides to choose or not to choose the currently seen element, based on the previously observed values, it cannot change its decision later regarding this element. The measure of success is the probability of selecting the highest value element, minimized over all adversarial assignments of values. We show existential and constructive upper bounds on approximation of the success probability in this problem, depending on the entropy of the randomly chosen arrival order, including the lowest possible entropy $O(\log\log (n))$ for which the probability of success could be constant. We show that below entropy level $\mathcal{H}<0.5\log\log n$, all algorithms succeed with probability $0$ if random order is selected uniformly at random from some subset of permutations, while we are able to construct in polynomial time a non-uniform distribution with entropy $\mathcal{H}$ resulting in success probability of at least $\Omega\left(\frac{1}{(\log\log n +3\log\log\log n -\mathcal{H})^{2+\epsilon}}\right)$, for any constant $\epsilon>0$. We also prove that no algorithm using entropy $\mathcal{H}=O((\log\log n)^a)$ can improve our result by more than polynomially, for any constant $0<a<1$. For entropy $\log\log (n)$ and larger, our analysis precisely quantifies both multiplicative and additive approximation of the success probability. In particular, we improve more than doubly exponentially on the best previously known additive approximation guarantee for the secretary problem.
Ambivalent and Consistent Relationships: The Role of Personal Networks in Cases of Domestic Violence
Social networks are usually considered as positive sources of social support, a role which has been extensively studied in the context of domestic violence. To victims of abuse, social networks often provide initial emotional and practical help as well useful information ahead of formal institutions. Recently, however, attention has been paid to the negative responses of social networks. In this article, we advance the theoretical debate on social networks as a source of social support by moving beyond the distinction between positive and negative ties. We do so by proposing the concepts of relational ambivalence and consistency, which describe the interactive processes by which people, intentionally or inadvertently, disregard or align with each other role relational expectations, therefore undermining or reinforcing individual choices of action. We analyse the qualitative accounts of nineteen female victims of domestic violence in Sweden, who described the responses of their personal networks during and after the abuse. We observe how the relationships embedded in these networks were described in ambivalent and consistent terms, and how they played a role in supporting or undermining women in reframing their loving relationships as abusive; in accounting or dismissing perpetrators responsibilities for the abuse; in relieving women from role expectations and obligations or in burdening them with further responsibilities; and in supporting or challenging their pathways out of domestic abuse. Our analysis suggests that social isolation cannot be considered a simple result of a lack of support but of the complex dynamics in which support is offered and accepted or withdrawn and refused.
This paper makes a first step towards a logic of learning from experiments. For this, we investigate formal frameworks for modeling the interaction of causal and (qualitative) epistemic reasoning. Crucial for our approach is the idea that the notion of an intervention can be used as a formal expression of a (real or hypothetical) experiment. In a first step we extend the well-known causal models with a simple Hintikka-style representation of the epistemic state of an agent. In the resulting setting, one can talk not only about the knowledge of an agent about the values of variables and how interventions affect them, but also about knowledge update. The resulting logic can model reasoning about thought experiments. However, it is unable to account for learning from experiments, which is clearly brought out by the fact that it validates the no learning principle for interventions. Therefore, in a second step, we implement a more complex notion of knowledge that allows an agent to observe (measure) certain variables when an experiment is carried out. This extended system does allow for learning from experiments. For all the proposed logical systems, we provide a sound and complete axiomatization.
This paper studies distributed binary test of statistical independence under communication (information bits) constraints. While testing independence is very relevant in various applications, distributed independence test is particularly useful for event detection in sensor networks where data correlation often occurs among observations of devices in the presence of a signal of interest. By focusing on the case of two devices because of their tractability, we begin by investigating conditions on Type I error probability restrictions under which the minimum Type II error admits an exponential behavior with the sample size. Then, we study the finite sample-size regime of this problem. We derive new upper and lower bounds for the gap between the minimum Type II error and its exponential approximation under different setups, including restrictions imposed on the vanishing Type I error probability. Our theoretical results shed light on the sample-size regimes at which approximations of the Type II error probability via error exponents became informative enough in the sense of predicting well the actual error probability. We finally discuss an application of our results where the gap is evaluated numerically, and we show that exponential approximations are not only tractable but also a valuable proxy for the Type II probability of error in the finite-length regime.
User interactions with recommender systems (RSs) are affected by user selection bias, e.g., users are more likely to rate popular items (popularity bias) or items that they expect to enjoy beforehand (positivity bias). Methods exist for mitigating the effects of selection bias in user ratings on the evaluation and optimization of RSs. However, these methods treat selection bias as static, despite the fact that the popularity of an item may change drastically over time and the fact that user preferences may also change over time. We focus on the age of an item and its effect on selection bias and user preferences. Our experimental analysis reveals that the rating behavior of users on the MovieLens dataset is better captured by methods that consider effects from the age of item on bias and preferences. We theoretically show that in a dynamic scenario in which both the selection bias and user preferences are dynamic, existing debiasing methods are no longer unbiased. To address this limitation, we introduce DebiAsing in the dyNamiC scEnaRio (DANCER), a novel debiasing method that extends the inverse propensity scoring debiasing method to account for dynamic selection bias and user preferences. Our experimental results indicate that DANCER improves rating prediction performance compared to debiasing methods that incorrectly assume that selection bias is static in a dynamic scenario. To the best of our knowledge, DANCER is the first debiasing method that accounts for dynamic selection bias and user preferences in RSs.
Iris recognition technology has attracted an increasing interest in the last decades in which we have witnessed a migration from research laboratories to real world applications. The deployment of this technology raises questions about the main vulnerabilities and security threats related to these systems. Among these threats presentation attacks stand out as some of the most relevant and studied. Presentation attacks can be defined as presentation of human characteristics or artifacts directly to the capture device of a biometric system trying to interfere its normal operation. In the case of the iris, these attacks include the use of real irises as well as artifacts with different level of sophistication such as photographs or videos. This chapter introduces iris Presentation Attack Detection (PAD) methods that have been developed to reduce the risk posed by presentation attacks. First, we summarise the most popular types of attacks including the main challenges to address. Secondly, we present a taxonomy of Presentation Attack Detection methods as a brief introduction to this very active research area. Finally, we discuss the integration of these methods into Iris Recognition Systems according to the most important scenarios of practical application.
We address the issue of tuning hyperparameters (HPs) for imitation learning algorithms in the context of continuous-control, when the underlying reward function of the demonstrating expert cannot be observed at any time. The vast literature in imitation learning mostly considers this reward function to be available for HP selection, but this is not a realistic setting. Indeed, would this reward function be available, it could then directly be used for policy training and imitation would not be necessary. To tackle this mostly ignored problem, we propose a number of possible proxies to the external reward. We evaluate them in an extensive empirical study (more than 10'000 agents across 9 environments) and make practical recommendations for selecting HPs. Our results show that while imitation learning algorithms are sensitive to HP choices, it is often possible to select good enough HPs through a proxy to the reward function.
Person re-identification (re-ID) has attracted much attention recently due to its great importance in video surveillance. In general, distance metrics used to identify two person images are expected to be robust under various appearance changes. However, our work observes the extreme vulnerability of existing distance metrics to adversarial examples, generated by simply adding human-imperceptible perturbations to person images. Hence, the security danger is dramatically increased when deploying commercial re-ID systems in video surveillance, especially considering the highly strict requirement of public safety. Although adversarial examples have been extensively applied for classification analysis, it is rarely studied in metric analysis like person re-identification. The most likely reason is the natural gap between the training and testing of re-ID networks, that is, the predictions of a re-ID network cannot be directly used during testing without an effective metric. In this work, we bridge the gap by proposing Adversarial Metric Attack, a parallel methodology to adversarial classification attacks, which can effectively generate adversarial examples for re-ID. Comprehensive experiments clearly reveal the adversarial effects in re-ID systems. Moreover, by benchmarking various adversarial settings, we expect that our work can facilitate the development of robust feature learning with the experimental conclusions we have drawn.
Methods that align distributions by minimizing an adversarial distance between them have recently achieved impressive results. However, these approaches are difficult to optimize with gradient descent and they often do not converge well without careful hyperparameter tuning and proper initialization. We investigate whether turning the adversarial min-max problem into an optimization problem by replacing the maximization part with its dual improves the quality of the resulting alignment and explore its connections to Maximum Mean Discrepancy. Our empirical results suggest that using the dual formulation for the restricted family of linear discriminators results in a more stable convergence to a desirable solution when compared with the performance of a primal min-max GAN-like objective and an MMD objective under the same restrictions. We test our hypothesis on the problem of aligning two synthetic point clouds on a plane and on a real-image domain adaptation problem on digits. In both cases, the dual formulation yields an iterative procedure that gives more stable and monotonic improvement over time.
小貼士
登錄享
注冊地址： 北京市海淀區羊坊店路18號2幢3層301-191