We applied a variety of parametric and non-parametric machine learning models to predict the probability distribution of rainfall based on 1M training examples over a single year across several U.S. states. Our top performing model based on a squared loss objective was a cross-validated parametric k-nearest-neighbor predictor that took about six days to compute, and was competitive in a world-w...

In this paper, we propose an approach to analyze the performance and the added value of automatic recommender systems in an industrial context. We show that recommender systems are multifaceted and can be organized around 4 structuring functions: help users to decide, help users to compare, help users to discover, help users to explore. A global off line protocol is then proposed to evaluate re...

Article history: Newmodel-based estimator Received 27 November 2007 Received in revised form 10 April 2008 Accepted 12 April 2008

K-means is a widely used iterative clustering algorithm. There has been considerable work on improving k-means in terms of mean squared error (MSE) and speed, both. However, most of the k-means variants tend to compute distance of each data point to each cluster centroid for every iteration. We propose two heuristics to overcome this bottleneck and speed up k-means. Our first heuristic predicts...

T HE THEORY developed in the statistical literature for the method of k-means can be applied to the study of optimal k-level vector quantizers. In this paper, I describe some of this theory, including a consistency theorem (Section II) and a central lim it theorem (Section IV) for k-means cluster centers. These results help to explain the behavior of optimal vector quantizers constructed from l...

Recently, a new pattern classiier using neighborhood information in the framework of the Dempster-Shafer theory of evidence was introduced 3, 2]. This approach consists in considering each neighbor of a pattern to be classiied as an item of evidence supporting certain hypotheses concerning the class membership of that pattern. In this paper, an adaptive version of this method is proposed, in wh...

Accurately mapping urban vegetation carbon density is challenging because of complex landscapes and mixed pixels. In this study, a novel methodology was proposed that combines a linear spectral unmixing analysis (LSUA) with a linear stepwise regression (LSR), a logistic model-based stepwise regression (LMSR) and k-Nearest Neighbors (kNN), to map the forest carbon density of Shenzhen City of Chi...

In addition to finding meaningful clusters, centroid-based clustering algorithms such as K-means or mean-shift should ideally find centroids that are valid patterns in the input space, representative of data in their cluster. This is challenging with data having a nonconvex or manifold structure, as with images or text. We introduce a new algorithm, Laplacian K-modes, which naturally combines t...

We present a novel unsupervised algorithm for quickly finding clusters in multidimensional data. It does not make the assumption of isotropy, instead taking full advantage of the anisotropic Gaussian kernel, to adapt to local data shape and scale. We employ some little-used properties of the multivariate Gaussian distribution to represent the data, and also give, as a corollary of the theory we...