We develop and analyze a projected particle Langevin optimization method to learn the distribution in the Sch\"{o}nberg integral representation of the radial basis functions from training samples. More specifically, we characterize a distributionally robust optimization method with respect to the Wasserstein distance to optimize the distribution in the Sch\"{o}nberg integral representation. To provide theoretical performance guarantees, we analyze the scaling limits of a projected particle online (stochastic) optimization method in the mean-field regime. In particular, we prove that in the scaling limits, the empirical measure of the Langevin particles converges to the law of a reflected It\^{o} diffusion-drift process. Moreover, the drift is also a function of the law of the underlying process. Using It\^{o} lemma for semi-martingales and Grisanov's change of measure for the Wiener processes, we then derive a Mckean-Vlasov type partial differential equation (PDE) with Robin boundary conditions that describes the evolution of the empirical measure of the projected Langevin particles in the mean-field regime. In addition, we establish the existence and uniqueness of the steady-state solutions of the derived PDE in the weak sense. We apply our learning approach to train radial kernels in the kernel locally sensitive hash (LSH) functions, where the training data-set is generated via a $k$-mean clustering method on a small subset of data-base. We subsequently apply our kernel LSH with a trained kernel for image retrieval task on MNIST data-set, and demonstrate the efficacy of our kernel learning approach. We also apply our kernel learning approach in conjunction with the kernel support vector machines (SVMs) for classification of benchmark data-sets.

Cancers diagnosed early are often more responsive to treatment. Blood tests that detect molecular markers of cancer have successfully identified individuals already known to have the disease. Lennon et al. conducted an exploratory study that more closely reflects the way in which such blood tests would be used in the future. They evaluated the feasibility and safety of incorporating a multicancer blood test into the routine clinical care of 10,000 women with no history of cancer. Over a 12-month period, the blood test detected 26 cancers of different types. A combination of the blood test and positron emission tomography–computed tomography (PET-CT) imaging led to surgical removal of nine of these cancers. Use of the blood test did not result in a large number of futile follow-up procedures. Science , this issue p. [eabb9601][1] ### INTRODUCTION The goal of earlier cancer detection is to identify the disease at a stage when it can be effectively treated, thereby offering the patient a better chance of long-term survival. Adherence to screening modalities known to decrease cancer mortality such as colonoscopy, mammography, low-dose computed tomography, and Pap smears varies widely. Moreover, the majority of cancer types are diagnosed only when symptoms occur. Multicancer blood tests offer the exciting possibility of detecting many cancer types at a relatively early stage and in a minimally invasive manner. ### RATIONALE Evaluation of the feasibility and safety of multicancer blood testing requires prospective interventional studies. We designed such a study to answer four critical questions: (i) Can a multicancer blood test detect cancers not previously detected by other means? (ii) Can a positive test result lead to surgical intervention with curative intent? (iii) Can testing be incorporated into routine clinical care and not discourage patients from undergoing recommended screening tests such as mammography? (iv) Can testing be performed safely, without incurring a large number of unnecessary, invasive follow-up tests? ### RESULTS We evaluated a blood test that detects DNA mutations and protein biomarkers of cancer in a prospective, interventional study of 10,006 women who were 65 to 75 years old and who had no prior history of cancer. Positive blood tests were followed by diagnostic positron emission tomography–computed tomography (PET-CT), which served to independently confirm and precisely localize the site and extent of disease if present. The study design incorporated several features to maximize the safety of testing to the participants. Of the 10,006 enrollees, 9911 (99.1%) could be assessed with respect to the four questions posed above. (i) Detection: Of 96 cancers incident during the study period, 26 were first detected by blood testing and 24 additional cancers by conventional screening. Fifteen of the 26 patients in whom cancer was first detected by blood testing underwent PET-CT imaging, and 11 patients developed signs or symptoms of cancer after the blood test that led to imaging procedures other than PET-CT. The specificity and positive predictive value (PPV) of blood testing alone were 98.9% and 19.4%, respectively, and combined with PET-CT, the specificity and PPV increased to 99.6% and 28.3%. The blood test first detected 14 of 45 cancers (31%) in seven organs for which no standard-of-care screening test is available. (ii) Intervention: Of the 26 cancers first detected by blood testing, 17 (65%) had localized or regional disease. Of the 15 participants with positive blood tests as well as positive PET-CT scans, 9 (60%) underwent surgery with curative intent. (iii) Incorporation into clinical care: Blood testing could be combined with conventional screening, leading to detection of more than half of the total incident cancers observed during the study period. Blood testing did not deter participants from undergoing mammography, and surveys revealed that 99% of participants would join a similar, subsequent study if offered. (iv) Safety: 99% of participants did not require any follow-up of blood testing results, and only 0.22% underwent an unnecessary invasive diagnostic procedure as a result of a false-positive blood test. ### CONCLUSION A minimally invasive blood test in combination with PET-CT can safely detect and precisely localize several types of cancers in individuals not previously known to have cancer, in some cases enabling treatment with intent to cure. Further studies will be required to assess the clinical utility, risk-benefit ratio, and cost-effectiveness of such testing. ![Figure][2] Overview of cancers detected by blood testing. Twenty-six cancers (blue dots) in 10 organs were first detected by blood testing. The blue dots with the red halo represent 12 of the 26 cancers that were surgically treated with intent to cure. Nine of these 12 were detected by the combination of the blood test and PET-CT, with the remaining three identified by the blood test combined with another imaging modality. Cancer treatments are often more successful when the disease is detected early. We evaluated the feasibility and safety of multicancer blood testing coupled with positron emission tomography–computed tomography (PET-CT) imaging to detect cancer in a prospective, interventional study of 10,006 women not previously known to have cancer. Positive blood tests were independently confirmed by a diagnostic PET-CT, which also localized the cancer. Twenty-six cancers were detected by blood testing. Of these, 15 underwent PET-CT imaging and nine (60%) were surgically excised. Twenty-four additional cancers were detected by standard-of-care screening and 46 by neither approach. One percent of participants underwent PET-CT imaging based on false-positive blood tests, and 0.22% underwent a futile invasive diagnostic procedure. These data demonstrate that multicancer blood testing combined with PET-CT can be safely incorporated into routine clinical care, in some cases leading to surgery with intent to cure. [1]: /lookup/doi/10.1126/science.abb9601 [2]: pending:yes

It is often convenient to combine precision and recall into a single metric called the F1 score, in particular, if you need a simple way to compare two classifiers. The F1 score is the harmonic mean of precision and recall. The F1 score favors classifiers that have similar precision and recall. This is not always what you want: in some contexts, you mostly care about precision, and in other contexts, you really care about the recall. For example, if you trained a classifier to detect videos that are safe for kids, you would probably prefer a classifier that rejects many good videos (low recall) but keeps only safe ones (high precision), rather than a classifier that has a much higher recall but lets a few really bad videos show up in your product (in such cases, you may even want to add a human pipeline to check the classifier's video selection). On the other hand, suppose you train a classifier to detect shoplifters on surveillance images: it is probably fine if your classifier has only 30% precision as long as it has 99% recall (sure, the security guards will get a few false alerts, but almost all shoplifters will get caught).

Understanding performance and prioritizing resources for the maintenance of the drinking-water pipe network throughout its life-cycle is a key part of water asset management. Renovation of this vital network is generally hindered by the difficulty or impossibility to gain physical access to the pipes. We study a statistical and machine learning framework for the prediction of water pipe failures. We employ classical and modern classifiers for a short-term prediction and survival analysis to provide a broader perspective and long-term forecast, usually needed for the economic analysis of the renovation. To enrich these models, we introduce new predictors based on water distribution domain knowledge and employ a modern oversampling technique to remedy the high imbalance coming from the few failures observed each year. For our case study, we use a dataset containing the failure records of all pipes within the water distribution network in Barcelona, Spain. The results shed light on the effect of important risk factors, such as pipe geometry, age, material, and soil cover, among others, and can help utility managers conduct more informed predictive maintenance tasks.

Preterm labor is the leading cause of neonatal morbidity and mortality and has attracted research efforts from many scientific areas. The inter-relationship between uterine contraction and the underlying electrical activities makes uterine electrohysterogram (EHG) a promising direction for preterm detection and prediction. Due the scarcity of EHG signals, especially those of preterm patients, synthetic algorithms are applied to create artificial samples of preterm type in order to remove prediction bias towards term, at the expense of a reduction of the feature effectiveness in machine-learning based automatic preterm detecting. To address such problem, we quantify the effect of synthetic samples (balance coefficient) on features' effectiveness, and form a general performance metric by utilizing multiple feature scores with relevant weights that describe their contributions to class separation. Combined with the activation/inactivation functions that characterizes the effect of the abundance of training samples in term and preterm prediction precision, we obtain an optimal sample balance coefficient that compromise the effect of synthetic samples in removing bias towards the majority and the side-effect of reducing features' importance. Substantial improvement in prediction precision has been achieved through a set of numerical tests on public available TPEHG database, and it verifies the effectiveness of the proposed method.

Skin lesion border irregularity is considered an important clinical feature for the early diagnosis of melanoma, representing the B feature in the ABCD rule. In this article we propose an automated approach for skin lesion border irregularity detection. The approach involves extracting the skin lesion from the image, detecting the skin lesion border, measuring the border irregularity, training a Convolutional Neural Network and Gaussian naive Bayes ensemble, to the automatic detection of border irregularity, which results in an objective decision on whether the skin lesion border is considered regular or irregular. The approach achieves outstanding results, obtaining an accuracy, sensitivity, specificity, and F-score of 93.6%, 100%, 92.5% and 96.1%, respectively.

Furthermore, the rate of reported crimes is dependent on the victims or bystanders to self-report. We design and implement an end-to-end system for real-time Though there exist algorithms for fully automated action crime detection in low-light environments. Unlike Closed-recognition [1-5], many are not applied in real-time Circuit Television, which performs reactively, the Low-Light or low-light environments. The existing benchmark action Environment Neural Surveillance provides real time crime recognition datasets such as HMDB-51 [6] (Human Motion alerts. The system uses a low-light video feed processed DataBase), UCF-101 [7] (University of Central Florida), in real-time by an optical-flow network, spatial and temporal and Sports-1M [8] contain primarily daytime videos. UCF networks, and a Support Vector Machine to identify released the UCF-Crime dataset [9] for general anomaly shootings, assaults, and thefts. We create a low-light actionrecognition detection and recognizes 13 crime categories, including arrest, dataset, LENS-4, which will be publicly available.

While current research has shown the importance of Multi-parametric MRI (mpMRI) in diagnosing prostate cancer (PCa), further investigation is needed for how to incorporate the specific structures of the mpMRI data, such as the regional heterogeneity and between-voxel correlation within a subject. This paper proposes a machine learning-based method for improved voxel-wise PCa classification by taking into account the unique structures of the data. We propose a multi-resolution modeling approach to account for regional heterogeneity, where base learners trained locally at multiple resolutions are combined using the super learner, and account for between-voxel correlation by efficient spatial Gaussian kernel smoothing. The method is flexible in that the super learner framework allows implementation of any classifier as the base learner, and can be easily extended to classifying cancer into more sub-categories. We describe detailed classification algorithm for the binary PCa status, as well as the ordinal clinical significance of PCa for which a weighted likelihood approach is implemented to enhance the detection of the less prevalent cancer categories. We illustrate the advantages of the proposed approach over conventional modeling and machine learning approaches through simulations and application to in vivo data.

Streaming data are increasingly present in real-world applications such as sensor measurements, satellite data feed, stock market, and financial data. The main characteristics of these applications are the online arrival of data observations at high speed and the susceptibility to changes in the data distributions due to the dynamic nature of real environments. The data stream mining community still faces some primary challenges and difficulties related to the comparison and evaluation of new proposals, mainly due to the lack of publicly available non-stationary real-world datasets. The comparison of stream algorithms proposed in the literature is not an easy task, as authors do not always follow the same recommendations, experimental evaluation procedures, datasets, and assumptions. In this paper, we mitigate problems related to the choice of datasets in the experimental evaluation of stream classifiers and drift detectors. To that end, we propose a new public data repository for benchmarking stream algorithms with real-world data. This repository contains the most popular datasets from literature and new datasets related to a highly relevant public health problem that involves the recognition of disease vector insects using optical sensors. The main advantage of these new datasets is the prior knowledge of their characteristics and patterns of changes to evaluate new adaptive algorithm proposals adequately. We also present an in-depth discussion about the characteristics, reasons, and issues that lead to different types of changes in data distribution, as well as a critical review of common problems concerning the current benchmark datasets available in the literature.

ML-based predictive systems are increasingly used to support decisions with a critical impact on individuals' lives such as college admission, job hiring, child custody, criminal risk assessment, etc. As a result, fairness emerged as an important requirement to guarantee that predictive systems do not discriminate against specific individuals or entire sub-populations, in particular, minorities. Given the inherent subjectivity of viewing the concept of fairness, several notions of fairness have been introduced in the literature. This paper is a survey of fairness notions that, unlike other surveys in the literature, addresses the question of "which notion of fairness is most suited to a given real-world scenario and why?". Our attempt to answer this question consists in (1) identifying the set of fairness-related characteristics of the real-world scenario at hand, (2) analyzing the behavior of each fairness notion, and then (3) fitting these two elements to recommend the most suitable fairness notion in every specific setup. The results are summarized in a decision diagram that can be used by practitioners and policy makers to navigate the relatively large catalogue of fairness notions.