Many interesting machine learning problems are best posed by considering instances that are distributions, or sample sets drawn from distributions. Most previous work devoted to machine learning tasks with distributional inputs has done so through pairwise kernel evaluations between pdfs (or sample sets). While such an approach is fine for smaller datasets, the computation of an N × N Gram matrix is prohibitive in large datasets. Recent scalable estimators that work over pdfs have done so only with kernels that use Euclidean metrics, like the L 2 distance. However, there are a myriad of other useful metrics available, such as total variation, Hellinger distance, and the Jensen-Shannon divergence. This work develops the first random features for pdfs whose dot product approximates kernels using these non-Euclidean metrics. These random features allow estimators to scale to large datasets by working in a primal space, without computing large Gram matrices. We provide an analysis of the approximation error in using our proposed random features, and show empirically the quality of our approximation both in estimating a Gram matrix and in solving learning tasks in real-world and synthetic data.

Unlike human learning, machine learning often fails to handle changes between training (source) and test (target) input distributions. Such domain shifts, common in practical scenarios, severely damage the performance of conventional machine learning methods. Supervised domain adaptation methods have been proposed for the case when the target data have labels, including some that perform very well despite being ``frustratingly easy'' to implement. However, in practice, the target domain is often unlabeled, requiring unsupervised adaptation. We propose a simple, effective, and efficient method for unsupervised domain adaptation called CORrelation ALignment (CORAL). CORAL minimizes domain shift by aligning the second-order statistics of source and target distributions, without requiring any target labels. Even though it is extraordinarily simple--it can be implemented in four lines of Matlab code--CORAL performs remarkably well in extensive evaluations on standard benchmark datasets.

A large amount of ordinal-valued data exist in many domains, including medical and health science, social science, economics, political science, etc. Unlike image and speech datasets of real-valued data, learning with ordinal variables (i.e., features) presents unique challenges. In particular, the nominal differences between those feature values, which are just ranks, do not necessarily correspond to the real distances between the corresponding categories. Given their wide existence, it is imperative to develop machine learning algorithms that specifically address the need to model and infer with such data. In this paper, we present a novel metric learning algorithm that takes into consideration the nature of ordinal data. Our approach treats ordinal values as latent variables in intervals. Our algorithm then learns what those intervals are as well as distance metrics to measure distances between latent variables in those intervals. We derive the corresponding optimization algorithm and demonstrate how that can be solved effectively. Experimental results show that the proposed approach significantly improves baselines that do not explicitly model ordinal features.

We study a novel machine learning (ML) problem setting of sequentially allocating small subsets of training data amongst a large set of classifiers. The goal is to select a classifier that will give near-optimal accuracy when trained on all data, while also minimizing the cost of misallocated samples. This is motivated by large modern datasets and ML toolkits with many combinations of learning algorithms and hyper-parameters. Inspired by the principle of "optimism under uncertainty," we propose an innovative strategy, Data Allocation using Upper Bounds (DAUB), which robustly achieves these objectives across a variety of real-world datasets. We further develop substantial theoretical support for DAUB in an idealized setting where the expected accuracy of a classifier trained on $n$ samples can be known exactly. Under these conditions we establish a rigorous sub-linear bound on the regret of the approach (in terms of misallocated data), as well as a rigorous bound on suboptimality of the selected classifier. Our accuracy estimates using real-world datasets only entail mild violations of the theoretical scenario, suggesting that the practical behavior of DAUB is likely to approach the idealized behavior.

Simultaneous optimistic optimization (SOO) is a recently proposed global optimization method with a strong theoretical foundation. Previous studies have shown that SOO has a good performance in low-dimensional optimization problems, however, its performance is unsatisfactory when the dimensionality is high. This paper adapts random embedding to scaling SOO, resulting in the RESOO algorithm. We prove that the simple regret of RESOO depends only on the effective dimension of the problem, while that of SOO depends on the dimension of the solution space. Empirically, on some high-dimensional non-convex testing functions as well as hyper-parameter tuning tasks for multi-class support vector machines, RESOO shows significantly improved performance from SOO.

Cluster validation constitutes one of the most challenging problems in unsupervised cluster analysis. For example, identifying the true number of clusters present in a dataset has been investigated for decades, and is still puzzling researchers today. The difficulty stems from the high variety of the dataset characteristics. Some datasets exhibit a strong structure with a few well-separated and normally distributed clusters, but most often real-world datasets contain possibly many overlapping non-gaussian clusters with heterogeneous variances and shapes. This calls for the design of robust clustering algorithms that could adapt to the structure of the data and in particular accurately guess the true number of clusters. They have recently been interesting attempts to design such algorithms, e.g. based on involved non-parametric statistical inference techniques. In this paper, we develop Viral Clustering (VC), a simple algorithm that jointly estimates the number of clusters and outputs clusters. The VC algorithm relies on two antagonist and interacting components. The first component tends to regroup neighbouring samples together, while the second component tends to spread samples in various clusters. This spreading component is performed using an analogy with the way virus spread over networks. We present extensive numerical experiments illustrating the robustness of the VC algorithm, and its superiority compared to existing algorithms.

Graph-based clustering methods perform clustering on a fixed input data graph. If this initial construction is of low quality then the resulting clustering may also be of low quality. Moreover, existing graph-based clustering methods require post-processing on the data graph to extract the clustering indicators. We address both of these drawbacks by allowing the data graph itself to be adjusted as part of the clustering procedure. In particular, our Constrained Laplacian Rank (CLR) method learns a graph with exactly k connected components (where k is the number of clusters). We develop two versions of this method, based upon the L1-norm and the L2-norm, which yield two new graph-based clustering objectives. We derive optimization algorithms to solve these objectives. Experimental results on synthetic datasets and real-world benchmark datasets exhibit the effectiveness of this new graph-based clustering method.

In recent data mining research, the graph clustering methods, such as normalized cut and ratio cut, have been well studied and applied to solve many unsupervised learning applications. The original graph clustering methods are NP-hard problems. Traditional approaches used spectral relaxation to solve the graph clustering problems. The main disadvantage of these approaches is that the obtained spectral solutions could severely deviate from the true solution. To solve this problem, in this paper, we propose a new relaxation mechanism for graph clustering methods. Instead of minimizing the squared distances of clustering results, we use the l1-norm distance. More important, considering the normalized consistency, we also use the l1-norm for the normalized terms in the new graph clustering relaxations. Due to the sparse result from the l1-norm minimization, the solutions of our new relaxed graph clustering methods get discrete values with many zeros, which are close to the ideal solutions. Our new objectives are difficult to be optimized, because the minimization problem involves the ratio of nonsmooth terms. The existing sparse learning optimization algorithms cannot be applied to solve this problem. In this paper, we propose a new optimization algorithm to solve this difficult non-smooth ratio minimization problem. The extensive experiments have been performed on three two-way clustering and eight multi-way clustering benchmark data sets. All empirical results show that our new relaxation methods consistently enhance the normalized cut and ratio cut clustering results.

Learning embeddings of entities and relations is an efficient and versatile method to perform machine learning on relational data such as knowledge graphs. In this work, we propose holographic embeddings (HolE) to learn compositional vector space representations of entire knowledge graphs. The proposed method is related to holographic models of associative memory in that it employs circular correlation to create compositional representations. By using correlation as the compositional operator, HolE can capture rich interactions but simultaneously remains efficient to compute, easy to train, and scalable to very large datasets. Experimentally, we show that holographic embeddings are able to outperform state-of-the-art methods for link prediction on knowledge graphs and relational learning benchmark datasets.

Conventional multi-label classification algorithms treat the target labels of the classification task as mere symbols that are void of an inherent semantics. However, in many cases textual descriptions of these labels are available or can be easily constructed from public document sources such as Wikipedia. In this paper, we investigate an approach for embedding documents and labels into a joint space while sharing word representations between documents and labels. For finding such embeddings, we rely on the text of documents as well as descriptions for the labels. The use of such label descriptions not only lets us expect an increased performance on conventional multi-label text classification tasks, but can also be used to make predictions for labels that have not been seen during the training phase. The potential of our method is demonstrated on the multi-label classification task of assigning keywords from the Medical Subject Headings (MeSH) to publications in biomedical research, both in a conventional and in a zero-shot learning setting.