Machine Learning explainability techniques have been proposed as a means of `explaining' or interrogating a model in order to understand why a particular decision or prediction has been made. Such an ability is especially important at a time when machine learning is being used to automate decision processes which concern sensitive factors and legal outcomes. Indeed, it is even a requirement according to EU law. Furthermore, researchers concerned with imposing overly restrictive functional form (e.g., as would be the case in a linear regression) may be motivated to use machine learning algorithms in conjunction with explainability techniques, as part of exploratory research, with the goal of identifying important variables which are associated with an outcome of interest. For example, epidemiologists might be interested in identifying `risk factors' - i.e. factors which affect recovery from disease - by using random forests and assessing variable relevance using importance measures. However, and as we demonstrate, machine learning algorithms are not as flexible as they might seem, and are instead incredibly sensitive to the underling causal structure in the data. The consequences of this are that predictors which are, in fact, critical to a causal system and highly correlated with the outcome, may nonetheless be deemed by explainability techniques to be unrelated/unimportant/unpredictive of the outcome. Rather than this being a limitation of explainability techniques per se, we show that it is rather a consequence of the mathematical implications of regression, and the interaction of these implications with the associated conditional independencies of the underlying causal structure. We provide some alternative recommendations for researchers wanting to explore the data for important variables.

Passenger flow allows the study of users' behavior through the public network and assists in designing new facilities and services. This flow is observed through interactions between passengers and infrastructure. For this task, Bluetooth technology and smartphones represent the ideal solution. The latter component allows users' identification, authentication, and billing, while the former allows short-range implicit interactions, device-to-device. To assess the potential of such a use case, we need to verify how robust Bluetooth signal and related machine learning (ML) classifiers are against the noise of realistic contexts. Therefore, we model binary passenger states with respect to a public vehicle, where one can either be-in or be-out (BIBO). The BIBO label identifies a fundamental building block of continuously-valued passenger flow. This paper describes the Human-Computer interaction experimental setting in a semi-controlled environment, which involves: two autonomous vehicles operating on two routes, serving three bus stops and eighteen users, as well as a proprietary smartphone-Bluetooth sensing platform. The resulting dataset includes multiple sensors' measurements of the same event and two ground-truth levels, the first being validation by participants, the second by three video-cameras surveilling buses and track. We performed a Monte-Carlo simulation of labels-flip to emulate human errors in the labeling process, as is known to happen in smartphone surveys; next we used such flipped labels for supervised training of ML classifiers. The impact of errors on model performance bias can be large. Results show ML tolerance to label flips caused by human or machine errors up to 30%.

We present an end-to-end differentiable neural network architecture to perform anomaly detection in multivariate time series by incorporating a Sequential Probability Ratio Test on the prediction residual. The architecture is a cascade of dynamical systems designed to separate linearly predictable components of the signal such as trends and seasonality, from the non-linear ones. The former are modeled by local Linear Dynamic Layers, and their residual is fed to a generic Temporal Convolutional Network that also aggregates global statistics from different time series as context for the local predictions of each one. The last layer implements the anomaly detector, which exploits the temporal structure of the prediction residuals to detect both isolated point anomalies and set-point changes. It is based on a novel application of the classic CUMSUM algorithm, adapted through the use of a variational approximation of f-divergences. The model automatically adapts to the time scales of the observed signals. It approximates a SARIMA model at the get-go, and auto-tunes to the statistics of the signal and its covariates, without the need for supervision, as more data is observed. The resulting system, which we call STRIC, outperforms both state-of-the-art robust statistical methods and deep neural network architectures on multiple anomaly detection benchmarks.

Naive Bayes(NB) is one of the essential algorithms in data mining. However, it is rarely used in reality because of the attribute independent assumption. Researchers have proposed many improved NB methods to alleviate this assumption. Among these methods, due to high efficiency and easy implementation, the filter attribute weighted NB methods receive great attentions. However, there still exists several challenges, such as the poor representation ability for single index and the fusion problem of two indexes. To overcome above challenges, we propose a general framework for Adaptive Two-index Fusion attribute weighted NB(ATFNB). Two types of data description category are used to represent the correlation between classes and attributes, intercorrelation between attributes and attributes, respectively. ATFNB can select any one index from each category. Then, we introduce a switching factor \{beta} to fuse two indexes, which can adaptively adjust the optimal ratio of the two index on various datasets. And a quick algorithm is proposed to infer the optimal interval of switching factor \{beta}. Finally, the weight of each attribute is calculated using the optimal value \{beta} and is integrated into NB classifier to improve the accuracy. The experimental results on 50 benchmark datasets and a Flavia dataset show that ATFNB outperforms the basic NB and state-of-the-art filter weighted NB models. In addition, the ATFNB framework can improve the existing two-index NB model by introducing the adaptive switching factor \{beta}. Auxiliary experimental results demonstrate the improved model significantly increases the accuracy compared to the original model without the adaptive switching factor \{beta}.

Fairness of machine learning algorithms has been of increasing interest. In order to suppress or eliminate discrimination in prediction, various notions as well as approaches have been proposed to impose fairness. Given a notion of fairness, an essential problem is then whether or not it can always be attained, even if with an unlimited amount of data. This issue is, however, not well addressed yet. In this paper, focusing on the Equalized Odds notion of fairness, we consider the attainability of this criterion and, furthermore, if it is attainable, the optimality of the prediction performance under various settings. In particular, for prediction performed by a deterministic function of input features, we give conditions under which Equalized Odds can hold true; if the stochastic prediction is acceptable, we show that under mild assumptions, fair predictors can always be derived. For classification, we further prove that compared to enforcing fairness by post-processing, one can always benefit from exploiting all available features during training and get potentially better prediction performance while remaining fair. Moreover, while stochastic prediction can attain Equalized Odds with theoretical guarantees, we also discuss its limitation and potential negative social impacts.

In some studies \citep[e.g.,][]{zhang2016understanding} of deep learning, it is observed that over-parametrized deep neural networks achieve a small testing error even when the training error is almost zero. Despite numerous works towards understanding this so-called "double descent" phenomenon \citep[e.g.,][]{belkin2018reconciling,belkin2019two}, in this paper, we turn into another way to enforce zero training error (without over-parametrization) through a data interpolation mechanism. Specifically, we consider a class of interpolated weighting schemes in the nearest neighbors (NN) algorithms. By carefully characterizing the multiplicative constant in the statistical risk, we reveal a U-shaped performance curve for the level of data interpolation in both classification and regression setups. This sharpens the existing result \citep{belkin2018does} that zero training error does not necessarily jeopardize predictive performances and claims a counter-intuitive result that a mild degree of data interpolation actually {\em strictly} improve the prediction performance and statistical stability over those of the (un-interpolated) $k$-NN algorithm. In the end, the universality of our results, such as change of distance measure and corrupted testing data, will also be discussed.

You create an amazing machine learning algorithm. You take a novel approach and apply techniques that prove to be highly accurate. Your results demonstrate a very high true positive rate and a very low false positive rate. You write a paper that articulates your outstanding results and submit it to a leading academic conference. You expect that this research will be well received, and you will receive many citations of your work.

Now, we trust the complex processes underlying artificial intelligence (AI) with everything from navigation to movie recommendations to targeted advertising. Can we also trust machine learning with our health care? The integration of AI and cancer care was a popular topic in 2021, as evidenced by prominent sessions at two of last year's AACR conferences: the 14th AACR Conference on the Science of Cancer Health Disparities in Racial/Ethnic Minorities and the Medically Underserved, held virtually October 6-8, 2021, and the San Antonio Breast Cancer Symposium (SABCS), held in a hybrid format December 7-10, 2021. During these sessions, experts gave an overview of how machine learning works, shared data on new applications of AI technologies, and emphasized important considerations for making algorithms equitable. Recognizing that a diverse audience of breast cancer clinicians and researchers may have questions about the fundamentals of AI, the SABCS session "Artificial Intelligence: Beyond the Soundbites" opened with a talk titled, "Everything You Always Wanted to Know About AI But Were Afraid to Ask," presented by Regina Barzilay, PhD, the AI faculty lead at the Jameel Clinic of the Massachusetts Institute of Technology.

Life assurance companies typically possess a wealth of data covering multiple systems and databases. These data are often used for analyzing the past and for describing the present. Taking account of the past, the future is mostly forecasted by traditional statistical methods. So far, only a few attempts were undertaken to perform estimations by means of machine learning approaches. In this work, the individual contract cancellation behavior of customers within two partial stocks is modeled by the aid of various classification methods. Partial stocks of private pension and endowment policy are considered. We describe the data used for the modeling, their structured and in which way they are cleansed. The utilized models are calibrated on the basis of an extensive tuning process, then graphically evaluated regarding their goodness-of-fit and with the help of a variable relevance concept, we investigate which features notably affect the individual contract cancellation behavior.

Augmented and Mixed Reality are emerging as likely successors to the mobile internet. However, many technical challenges remain. One of the key requirements of these systems is the ability to create a continuity between physical and virtual worlds, with the user's visual perception as the primary interface medium. Building this continuity requires the system to develop a visual understanding of the physical world. While there has been significant recent progress in computer vision and AI techniques such as image classification and object detection, success in these areas has not yet led to the visual perception required for these critical MR and AR applications. A significant issue is that current evaluation criteria are insufficient for these applications. To motivate and evaluate progress in this emerging area, there is a need for new metrics. In this paper we outline limitations of current evaluation criteria and propose new criteria.