Relational probabilistic models have the challenge of aggregation, where one variable depends on a population of other variables. Consider the problem of predicting gender from movie ratings; this is challenging because the number of movies per user and users per movie can vary greatly. Surprisingly, aggregation is not well understood. In this paper, we show that existing relational models (implicitly or explicitly) either use simple numerical aggregators that lose great amounts of information, or correspond to naive Bayes, logistic regression, or noisy-OR that suffer from overconfidence. We propose new simple aggregators and simple modifications of existing models that empirically outperform the existing ones. The intuition we provide on different (existing or new) models and their shortcomings plus our empirical findings promise to form the foundation for future representations.

We improve existing instability-based methods for the selection of the number of clusters $k$ in cluster analysis by normalizing instability. In contrast to existing instability methods which only perform well for bounded sequences of small $k$, our method performs well across the whole sequence of possible $k$. In addition, we compare for the first time model-based and model-free variants of $k$ selection via cluster instability and find that their performance is similar. We make our method available in the R-package \verb+cstab+.

A genome-wide association study (GWAS) correlates marker variation with trait variation in a sample of individuals. Each study subject is genotyped at a multitude of SNPs (single nucleotide polymorphisms) spanning the genome. Here we assume that subjects are unrelated and collected at random and that trait values are normally distributed or transformed to normality. Over the past decade, researchers have been remarkably successful in applying GWAS analysis to hundreds of traits. The massive amount of data produced in these studies present unique computational challenges. Penalized regression with LASSO or MCP penalties is capable of selecting a handful of associated SNPs from millions of potential SNPs. Unfortunately, model selection can be corrupted by false positives and false negatives, obscuring the genetic underpinning of a trait. This paper introduces the iterative hard thresholding (IHT) algorithm to the GWAS analysis of continuous traits. Our parallel implementation of IHT accommodates SNP genotype compression and exploits multiple CPU cores and graphics processing units (GPUs). This allows statistical geneticists to leverage commodity desktop computers in GWAS analysis and to avoid supercomputing. We evaluate IHT performance on both simulated and real GWAS data and conclude that it reduces false positive and false negative rates while remaining competitive in computational time with penalized regression. Source code is freely available at https://github.com/klkeys/IHT.jl.

Human-centered environments are rich with a wide variety of spatial relations between everyday objects. For autonomous robots to operate effectively in such environments, they should be able to reason about these relations and generalize them to objects with different shapes and sizes. For example, having learned to place a toy inside a basket, a robot should be able to generalize this concept using a spoon and a cup. This requires a robot to have the flexibility to learn arbitrary relations in a lifelong manner, making it challenging for an expert to pre-program it with sufficient knowledge to do so beforehand. In this paper, we address the problem of learning spatial relations by introducing a novel method from the perspective of distance metric learning. Our approach enables a robot to reason about the similarity between pairwise spatial relations, thereby enabling it to use its previous knowledge when presented with a new relation to imitate. We show how this makes it possible to learn arbitrary spatial relations from non-expert users using a small number of examples and in an interactive manner. Our extensive evaluation with real-world data demonstrates the effectiveness of our method in reasoning about a continuous spectrum of spatial relations and generalizing them to new objects.

Evaluating expression of the Human epidermal growth factor receptor 2 (Her2) by visual examination of immunohistochemistry (IHC) on invasive breast cancer (BCa) is a key part of the diagnostic assessment of BCa due to its recognised importance as a predictive and prognostic marker in clinical practice. However, visual scoring of Her2 is subjective and consequently prone to inter-observer variability. Given the prognostic and therapeutic implications of Her2 scoring, a more objective method is required. In this paper, we report on a recent automated Her2 scoring contest, held in conjunction with the annual PathSoc meeting held in Nottingham in June 2016, aimed at systematically comparing and advancing the state-of-the-art Artificial Intelligence (AI) based automated methods for Her2 scoring. The contest dataset comprised of digitised whole slide images (WSI) of sections from 86 cases of invasive breast carcinoma stained with both Haematoxylin & Eosin (H&E) and IHC for Her2. The contesting algorithms automatically predicted scores of the IHC slides for an unseen subset of the dataset and the predicted scores were compared with the 'ground truth' (a consensus score from at least two experts). We also report on a simple Man vs Machine contest for the scoring of Her2 and show that the automated methods could beat the pathology experts on this contest dataset. This paper presents a benchmark for comparing the performance of automated algorithms for scoring of Her2. It also demonstrates the enormous potential of automated algorithms in assisting the pathologist with objective IHC scoring.

H2O's Driverless AI promises to bring ML analysis to nontechnical users, and to take the drudgery out of model selection for experts H2O.ai, creator of applications for making machine learning accessible to business users, has introduced a product intended to allow business users familiar with products like Tableau to extract insights from data without needing expertise in deploying or tuning machine learning models. Driverless AI, currently in beta, is billed by H2O.ai as an "expert system for AI" -- a way to automate the kinds of expertise that data scientists bring to developing machine learning models. The target audience is non-expert users, who can take datasets and run GPU-accelerated ML algorithms against them to extract useful results, without understanding the ins and outs of data science. In addition to business users eager to leverage ML in their organizations but lack expertise, H2O is also pitching Driverless AI to data scientists. H2O considers Driverless AI to be a way for expert users to automate some of the more tedious processes of analyzing a dataset, such as selecting which of various automatically trained models is the best fit for a given dataset. The end user sets up their data experiments by way of a web-based UI, with the user typically needing only to choose which target variable from the dataset to solve for.

IBM Data Science Experience (DSX) is an interactive, collaborative, cloud-based environment where data scientists can use multiple tools to achieve insights. Data scientists can use the best of open source, tap into IBM's unique features, grow their skills, and collaborate with teams. One of the many features of DSX provides the capability to create and train a machine learning model in DSX with little to no coding. This model can subsequently be saved and deployed to Watson Machine Learning on IBM Bluemix and called for scoring in real-time. This tutorial is a continuation of the following logistic regression analysis, which creates, trains, saves and deploys a logistic regression model that predicts the possibility for a tent purchase based on age, sex, marital status, and job profession for an individual.

Kernel methods are a group of machine learning algorithms used for pattern analysis, which involves organizing raw data into rankings, clusters, or classifications. These methods allow data scientists to apply their domain knowledge of a given problem by building custom kernels that incorporate the data transformations that are most likely to improve the accuracy of the overall mode The most popular application of kernels is the support vector machine (SVM), which builds a model that classifies new data as belonging to one category or another based on a set of training examples. A SVM makes these determinations by representing each example as a point in a multi-dimensional space called a hyperplane. The points are then separated into categories by maximizing the distance (called a "margin") between the different apparent groups in the data.

The median heuristic is a popular tool to set the bandwidth of radial basis function kernels. While its empirical performances make it a safe choice under most circumstances, there is little theoretical understanding of why this is the case. For large sample size, we show in this article that the median heuristic behaves approximately as the median of a distribution that we describe completely in the setting of kernel two-sample test and kernel change-point detection. More precisely, we show that the median heuristic is asymptotically normal around this value. We illustrate these findings when the underlying distributions are multivariate Gaussian distributions.

Embeddings of knowledge graphs have received significant attention due to their excellent performance for tasks like link prediction and entity resolution. In this short paper, we are providing a comparison of two state-of-the-art knowledge graph embeddings for which their equivalence has recently been established, i.e., ComplEx and HolE [Nickel, Rosasco, and Poggio, 2016; Trouillon et al., 2016; Hayashi and Shimbo, 2017]. First, we briefly review both models and discuss how their scoring functions are equivalent. We then analyze the discrepancy of results reported in the original articles, and show experimentally that they are likely due to the use of different loss functions. In further experiments, we evaluate the ability of both models to embed symmetric and antisymmetric patterns. Finally, we discuss advantages and disadvantages of both models and under which conditions one would be preferable to the other.