Non-Intrusive Load Monitoring (NILM) aims to predict the status or consumption of domestic appliances in a household only by knowing the aggregated power load. NILM can be formulated as regression problem or most often as a classification problem. Most datasets gathered by smart meters allow to define naturally a regression problem, but the corresponding classification problem is a derived one, since it requires a conversion from the power signal to the status of each device by a thresholding method. We treat three different thresholding methods to perform this task, discussing their differences on various devices from the UK-DALE dataset. We analyze the performance of deep learning state-of-the-art architectures on both the regression and classification problems, introducing criteria to select the most convenient thresholding method.

The environments in which we deploy machine learning (ML) algorithms rarely look exactly like the environments in which we collected our training data. Unfortunately, we lack methodology for evaluating how well an algorithm will generalize to new environments that differ in a structured way from the training data (i.e., the case of dataset shift (Quiñonero-Candela et al., 2009)). Such methodology is increasingly important as ML systems are being deployed across a number of industries, such as health care and personal finance, in which system performance translates directly to real-world outcomes. Further, as regulation and product reviews become more common across industries, system developers will be expected to produce evidence of the validity and safety of their systems. For example, the United States Food and Drug Administration (FDA) currently regulates ML systems for medical applications, requiring evidence for the validity of such systems before approval is granted (US Food and Drug Administration, 2019). Evaluation methods for assessing model validity have typically focused on how the model performs on data from the training distribution, known as internal validity. Powerful tools, such as cross-validation and the bootstrap, satisfy the assumption that the training and test data are drawn from the same distribution. However, these validation methods do not capture a model's ability to generalize to new environments, known as external validity (Campbell and Stanley, 1963). Currently, the main way to assess a model's external validity is to empirically evaluate performance on multiple, independently collected datasets (e.g.,

To date, there has been no formal study of the statistical cost of interpretability in machine learning. As such, the discourse around potential trade-offs is often informal and misconceptions abound. In this work, we aim to initiate a formal study of these trade-offs. A seemingly insurmountable roadblock is the lack of any agreed upon definition of interpretability. Instead, we propose a shift in perspective. Rather than attempt to define interpretability, we propose to model the \emph{act} of \emph{enforcing} interpretability. As a starting point, we focus on the setting of empirical risk minimization for binary classification, and view interpretability as a constraint placed on learning. That is, we assume we are given a subset of hypothesis that are deemed to be interpretable, possibly depending on the data distribution and other aspects of the context. We then model the act of enforcing interpretability as that of performing empirical risk minimization over the set of interpretable hypotheses. This model allows us to reason about the statistical implications of enforcing interpretability, using known results in statistical learning theory. Focusing on accuracy, we perform a case analysis, explaining why one may or may not observe a trade-off between accuracy and interpretability when the restriction to interpretable classifiers does or does not come at the cost of some excess statistical risk. We close with some worked examples and some open problems, which we hope will spur further theoretical development around the tradeoffs involved in interpretability.

Deep learning methods for graphs achieve remarkable performance across a variety of domains. However, recent findings indicate that small, unnoticeable perturbations of graph structure can catastrophically reduce performance of even the strongest and most popular Graph Neural Networks (GNNs). Here, we develop GNNGuard, a general algorithm to defend against a variety of training-time attacks that perturb the discrete graph structure. GNNGuard can be straight-forwardly incorporated into any GNN. Its core principle is to detect and quantify the relationship between the graph structure and node features, if one exists, and then exploit that relationship to mitigate negative effects of the attack.GNNGuard learns how to best assign higher weights to edges connecting similar nodes while pruning edges between unrelated nodes. The revised edges allow for robust propagation of neural messages in the underlying GNN. GNNGuard introduces two novel components, the neighbor importance estimation, and the layer-wise graph memory, and we show empirically that both components are necessary for a successful defense. Across five GNNs, three defense methods, and five datasets,including a challenging human disease graph, experiments show that GNNGuard outperforms existing defense approaches by 15.3% on average. Remarkably, GNNGuard can effectively restore state-of-the-art performance of GNNs in the face of various adversarial attacks, including targeted and non-targeted attacks, and can defend against attacks on heterophily graphs.

The hypothesis that high-dimensional data tend to live near a manifold of low dimension is an important theme of modern statistics and machine learning, often held to explain why highdimensional learning is realistically possible [61, 5, 15, 13, 18, 44]. The object of this paper is show that, for a theoretically tractable but rich class of random graph models, such a phenomenon occurs in the spectral embedding of a graph. Manifold structure is shown to arise when the graph follows a latent position model [28], wherein connections are posited to occur as a function of the nodes' underlying positions in space. Because of their intuitive appeal, such models have been employed in a great diversity of disciplines, including social science [35, 42, 21], neuroscience [17, 52], statistical mechanics [34], information technology [69], biology [53] and ecology [19]. In many more endeavours latent position models are used -- at least according to Definition 1 (to follow) -- but are known by a different name; examples include the standard [29], mixed [2] and degree-corrected [32] stochastic block models, random geometric graphs [50], and the graphon model [39], which encompasses them.

As automated vehicles receive more attention from the media, there has been an equivalent increase in the coverage of the ethical choices a vehicle may be forced to make in certain crash situations with no clear safe outcome. Much of this coverage has focused on a philosophical thought experiment known as the "trolley problem," and substituting an automated vehicle for the trolley and the car's software for the bystander. While this is a stark and straightforward example of ethical decision making for an automated vehicle, it risks marginalizing the entire field if it is to become the only ethical problem in the public's mind. In this chapter, I discuss the shortcomings of the trolley problem, and introduce more nuanced examples that involve crash risk and uncertainty. Risk management is introduced as an alternative approach, and its ethical dimensions are discussed.

Recent advances in causal inference techniques, more specifically, in the theory of structural causal models, provide the framework for identification of causal effects from observational data in the cases where the causal graph is identifiable, i.e., the data generating mechanism can be recovered from the joint distribution. However, no such studies have been done to demonstrate this concept with a clinical example. We present a complete framework to estimate the causal effect from observational data by augmenting expert knowledge in the model development phase and with a practical clinical application. Our clinical application entails a timely and important research question, i.e., the effect of oxygen therapy intervention in the intensive care unit (ICU); the result of this project is useful in a variety of disease conditions, including severe acute respiratory syndrome coronavirus-2 (SARS-CoV-2) patients in the ICU. We used data from the MIMIC III database, a standard database in the machine learning community that contains 58,976 admissions from an ICU in Boston, MA, for estimating the oxygen therapy effect on morality. We also identified the covariate-specific effect to oxygen therapy from the model for more personalized intervention.

One area artificial intelligence can have a deep impact on is climate predictions. Google already has a number of weather efforts, and this month launched an AI partnership with the National Oceanic and Atmospheric Administration (NOAA) Satellite and Information Service (NESDIS) to enhance analysis of the US agency's satellite and environmental data. Together, NESDIS and Google will use AI and ML to amplify NOAA's environmental monitoring, weather forecasting, and climate research using Google Cloud infrastructure. Both organizations will initially research small-scale AI/ML systems, before "executing full-scale prototypes that NOAA could ultimately operationalize across its organization." NOAA says this has the "potential to be a significant leap" in its ability to analyze an "enormous volume and diversity of environmental data" to ultimately "enhance prediction for extreme weather events," with hurricanes and tornadoes cited.

Drones of all sizes are being used by environmental advocates to monitor deforestation, by conservationists to track poachers, and by journalists and activists to document large protests. As a political sociologist who studies social movements and drones, I document a wide range of nonviolent and pro-social drone uses in my new book, "The Good Drone." I show that these efforts have the potential to democratize surveillance. But when the Department of Homeland Security redirects large, fixed-wing drones from the U.S.-Mexico border to monitor protests, and when towns experiment with using drones to test people for fevers, it's time to think about how many eyes are in the sky and how to avoid unwanted aerial surveillance. One way that's within reach of nearly everyone is learning how to simply disappear from view.