KNN is one of the most popular classification methods, but it often fails to work well with inappropriate choice of distance metric or due to the presence of numerous class-irrelevant features. Linear feature transformation methods have been widely applied to extract class-relevant information to improve kNN classification, which is very limited in many applications. Kernels have been used to learn powerful non-linear feature transformations, but these methods fail to scale to large datasets. In this paper, we present a scalable non-linear feature mapping method based on a deep neural network pretrained with restricted boltzmann machines for improving kNN classification in a large-margin framework, which we call DNet-kNN. DNet-kNN can be used for both classification and for supervised dimensionality reduction. The experimental results on two benchmark handwritten digit datasets show that DNet-kNN has much better performance than large-margin kNN using a linear mapping and kNN based on a deep autoencoder pretrained with retricted boltzmann machines.

We present a method for learning max-weight matching predictors in bipartite graphs. The method consists of performing maximum a posteriori estimation in exponential families with sufficient statistics that encode permutations and data features. Although inference is in general hard, we show that for one very relevant application - web page ranking - exact inference is efficient. For general model instances, an appropriate sampler is readily available. Contrary to existing max-margin matching models, our approach is statistically consistent and, in addition, experiments with increasing sample sizes indicate superior improvement over such models. We apply the method to graph matching in computer vision as well as to a standard benchmark dataset for learning web page ranking, in which we obtain state-of-the-art results, in particular improving on max-margin variants. The drawback of this method with respect to max-margin alternatives is its runtime for large graphs, which is comparatively high.

We consider the problem of estimating the conditional probability of a label in time $O(\log n)$, where $n$ is the number of possible labels. We analyze a natural reduction of this problem to a set of binary regression problems organized in a tree structure, proving a regret bound that scales with the depth of the tree. Motivated by this analysis, we propose the first online algorithm which provably constructs a logarithmic depth tree on the set of labels to solve this problem. We test the algorithm empirically, showing that it works succesfully on a dataset with roughly $10^6$ labels.

We present a novel approach for learning nonlinear dynamic models, which leads to a new set of tools capable of solving problems that are otherwise difficult. We provide theory showing this new approach is consistent for models with long range structure, and apply the approach to motion capture and high-dimensional video data, yielding results superior to standard alternatives.

Data analysis and data mining are concerned with unsupervised pattern finding and structure determination in data sets. The data sets themselves are explicitly linked as a form of representation to an observational or otherwise empirical domain of interest. "Structure" has long been understood as symmetry which can take many forms with respect to any transformation, including point, translational, rotational, and many others. Beginning with the role of number theory in expressing data, we show how we can naturally proceed to hierarchical structures. We show how this both encapsulates traditional paradigms in data analysis, and also opens up new perspectives towards issues that are on the order of the day, including data mining of massive, high dimensional, heterogeneous data sets. Linkages with other fields are also discussed including computational logic and symbolic dynamics. The structures in data surveyed here are based on hierarchy, represented as p-adic numbers or an ultrametric topology.

In this paper, we present an application of neural networks in the renewable energy domain. We have developed a methodology for the daily prediction of global solar radiation on a horizontal surface. We use an ad-hoc time series preprocessing and a Multi-Layer Perceptron (MLP) in order to predict solar radiation at daily horizon. First results are promising with nRMSE < 21% and RMSE < 998 Wh/m2. Our optimized MLP presents prediction similar to or even better than conventional methods such as ARIMA techniques, Bayesian inference, Markov chains and k-Nearest-Neighbors approximators. Moreover we found that our data preprocessing approach can reduce significantly forecasting errors.

Many regression problems involve not one but several response variables (y's). Often the responses are suspected to share a common underlying structure, in which case it may be advantageous to share information across them; this is known as multitask learning. As a special case, we can use multiple responses to better identify shared predictive features -- a project we might call multitask feature selection. This thesis is organized as follows. Section 1 introduces feature selection for regression, focusing on ell_0 regularization methods and their interpretation within a Minimum Description Length (MDL) framework. Section 2 proposes a novel extension of MDL feature selection to the multitask setting. The approach, called the "Multiple Inclusion Criterion" (MIC), is designed to borrow information across regression tasks by more easily selecting features that are associated with multiple responses. We show in experiments on synthetic and real biological data sets that MIC can reduce prediction error in settings where features are at least partially shared across responses. Section 3 surveys hypothesis testing by regression with a single response, focusing on the parallel between the standard Bonferroni correction and an MDL approach. Mirroring the ideas in Section 2, Section 4 proposes a novel MIC approach to hypothesis testing with multiple responses and shows that on synthetic data with significant sharing of features across responses, MIC sometimes outperforms standard FDR-controlling methods in terms of finding true positives for a given level of false positives. Section 5 concludes.

Complex questions that require inferencing and synthesizing information from multiple documents can be seen as a kind of topic-oriented, informative multi-document summarization where the goal is to produce a single text as a compressed version of a set of documents with a minimum loss of relevant information. In this paper, we experiment with one empirical method and two unsupervised statistical machine learning techniques: K-means and Expectation Maximization (EM), for computing relative importance of the sentences. We compare the results of these approaches. Our experiments show that the empirical approach outperforms the other two techniques and EM performs better than K-means. However, the performance of these approaches depends entirely on the feature set used and the weighting of these features. In order to measure the importance and relevance to the user query we extract different kinds of features (i.e. lexical, lexical semantic, cosine similarity, basic element, tree kernel based syntactic and shallow-semantic) for each of the document sentences. We use a local search technique to learn the weights of the features. To the best of our knowledge, no study has used tree kernel functions to encode syntactic/semantic information for more complex tasks such as computing the relatedness between the query sentences and the document sentences in order to generate query-focused summaries (or answers to complex questions). For each of our methods of generating summaries (i.e. empirical, K-means and EM) we show the effects of syntactic and shallow-semantic features over the bag-of-words (BOW) features.

Clustering and visualization of large text document collections aids in browsing, navigation, and information retrieval. We present a document clustering and visualization method based on Latent Dirichlet Allocation and self-organizing maps (LDA-SOM). LDA-SOM clusters documents based on topical content and renders clusters in an intuitive two-dimensional format. Document topics are inferred using a probabilistic topic model. Then, due to the topology preserving properties of self-organizing maps, document clusters with similar topic distributions are placed near one another in the visualization. This provides the user an intuitive means of browsing from one cluster to another based on topics held in common. The effectiveness of LDA-SOM is evaluated on the 20 Newsgroups and NIPS data sets.

In this paper, we are interested in the qualitative knowledge that underlies some given probabilistic information. To represent such qualitative structures, we use ordinal conditional functions, OCFs, (or ranking functions) as a qualitative abstraction of probability functions. The basic idea for transforming probabilities into ordinal rankings is to find well-behaved clusterings of the negative logarithms of the probabilities. We show how popular clustering tools can be used for this, and propose measures for the evaluation of the clustering results in this context. From the so obtained ranking functions, we extract conditionals that may serve as a base for inductive default reasoning.