Many studies in biomedical and health sciences involve small sample sizes due to logistic or financial constraints. Often, identifying weak (but scientifically interesting) associations between a set of predictors and a response necessitates pooling datasets from multiple diverse labs or groups. While there is a rich literature in statistical machine learning to address distributional shifts and inference in multi-site datasets, it is less clear ${\it when}$ such pooling is guaranteed to help (and when it does not) -- independent of the inference algorithms we use. In this paper, we present a hypothesis test to answer this question, both for classical and high dimensional linear regression. We precisely identify regimes where pooling datasets across multiple sites is sensible, and how such policy decisions can be made via simple checks executable on each site before any data transfer ever happens. With a focus on Alzheimer's disease studies, we present empirical results showing that in regimes suggested by our analysis, pooling a local dataset with data from an international study improves power.

Preprocessing data is an important step before any data analysis. In this paper, we focus on one particular aspect, namely scaling or normalization. We analyze various scaling methods in common use and study their effects on different statistical learning models. We will propose a new two-stage scaling method. First, we use some training data to fit linear regression model and then scale the whole data based on the coefficients of regression. Simulations are conducted to illustrate the advantages of our new scaling method. Some real data analysis will also be given.

In this letter, we propose a new identification criterion that guarantees the recovery of the low-rank latent factors in the nonnegative matrix factorization (NMF) model, under mild conditions. Specifically, using the proposed criterion, it suffices to identify the latent factors if the rows of one factor are \emph{sufficiently scattered} over the nonnegative orthant, while no structural assumption is imposed on the other factor except being full-rank. This is by far the mildest condition under which the latent factors are provably identifiable from the NMF model.

Categorical compositional distributional semantics is a model of natural language; it combines the statistical vector space models of words with the compositional models of grammar. We formalise in this model the generalised quantifier theory of natural language, due to Barwise and Cooper. The underlying setting is a compact closed category with bialgebras. We start from a generative grammar formalisation and develop an abstract categorical compositional semantics for it, then instantiate the abstract setting to sets and relations and to finite dimensional vector spaces and linear maps. We prove the equivalence of the relational instantiation to the truth theoretic semantics of generalised quantifiers. The vector space instantiation formalises the statistical usages of words and enables us to, for the first time, reason about quantified phrases and sentences compositionally in distributional semantics.

Check out Siddha Ganju's talk on embedded deep learning at the Artificial Intelligence Conference in San Francisco, Sept. 17-20, 2017. Learning rate is the rate at which the accumulation of information in a neural network progresses over time. The learning rate determines how quickly (and whether at all) the network reaches the optimum, most conducive location in the network for the specific output desired. In plain Stochastic Gradient Descent (SGD), the learning rate is not related to the shape of the error gradient because a global learning rate is used, which is independent of the error gradient. However, there are many modifications that can be made to the original SGD update rule that relates the learning rate to the magnitude and orientation of the error gradient.

Jaguar Land Rover, taking a page from the European luxury car playbook, is offering increasingly attractive performance versions of its entry-level sports cars. One of those was on display at Mazda Raceway Laguna Seca during the recent Monterey Car Week, which culminates at Pebble Beach with the famed Concours D'Elegance. At the track, days before the Concours, Jaguar designers showed off the XE SV Project 8, a street-legal track car, in front of the raceway pits. They boasted deservedly about its good looks and great specifications -- including its 592 horsepower and $187,500 sticker price. You don't have to buy one.

The model predicts or estimates price (target) as a function of engine size, horse power, and width (predictors). Recall that multivariate regression model assumes independence between the independent predictors. It treats horsepower, engine size, and width as if they are not related. In practice, variables are rarely independent. This blog post will address this question.

We propose a new algorithm, Mean Actor-Critic (MAC), for discrete-action continuous-state reinforcement learning. MAC is a policy gradient algorithm that uses the agent's explicit representation of all action values to estimate the gradient of the policy, rather than using only the actions that were actually executed. This significantly reduces variance in the gradient updates and removes the need for a variance reduction baseline. We show empirical results on two control domains where MAC performs as well as or better than other policy gradient approaches, and on five Atari games, where MAC is competitive with state-of-the-art policy search algorithms.

In this work, we address the problem of disentanglement of factors that generate a given data into those that are correlated with the labeling and those that are not. Our solution is simpler than previous solutions and employs adversarial training in a straightforward manner. We demonstrate the new method on visual datasets as well as on financial data. In order to evaluate the latter, we developed a hypothetical trading strategy whose performance is affected by the performance of the disentanglement, namely, it trades better when the factors are better separated.

This paper presents convergence analysis of kernel-based quadrature rules in misspecified settings, focusing on deterministic quadrature in Sobolev spaces. In particular, we deal with misspecified settings where a test integrand is less smooth than a Sobolev RKHS based on which a quadrature rule is constructed. We provide convergence guarantees based on two different assumptions on a quadrature rule: one on quadrature weights, and the other on design points. More precisely, we show that convergence rates can be derived (i) if the sum of absolute weights remains constant (or does not increase quickly), or (ii) if the minimum distance between distance design points does not decrease very quickly. As a consequence of the latter result, we derive a rate of convergence for Bayesian quadrature in misspecified settings. We reveal a condition on design points to make Bayesian quadrature robust to misspecification, and show that, under this condition, it may adaptively achieve the optimal rate of convergence in the Sobolev space of a lesser order (i.e., of the unknown smoothness of a test integrand), under a slightly stronger regularity condition on the integrand.