This paper introduces the kernel mixture network, a new method for nonparametric estimation of conditional probability densities using neural networks. We model arbitrarily complex conditional densities as linear combinations of a family of kernel functions centered at a subset of training points. The weights are determined by the outer layer of a deep neural network, trained by minimizing the negative log likelihood. This generalizes the popular quantized softmax approach, which can be seen as a kernel mixture network with square and non-overlapping kernels. We test the performance of our method on two important applications, namely Bayesian filtering and generative modeling. In the Bayesian filtering example, we show that the method can be used to filter complex nonlinear and non-Gaussian signals defined on manifolds. The resulting kernel mixture network filter outperforms both the quantized softmax filter and the extended Kalman filter in terms of model likelihood. Finally, our experiments on generative models show that, given the same architecture, the kernel mixture network leads to higher test set likelihood, less overfitting and more diversified and realistic generated samples than the quantized softmax approach.

The concepts of sketching and subsampling have recently received much attention by the optimization and statistics communities. In this paper, we study Newton-Sketch and Subsampled Newton (SSN) methods for the finite-sum optimization problem. We consider practical versions of the two methods in which the Newton equations are solved approximately using the conjugate gradient (CG) method or a stochastic gradient iteration. We establish new complexity results for the SSN-CG method that exploit the spectral properties of CG. Controlled numerical experiments compare the relative strengths of Newton-Sketch and SSN methods and show that for many finite-sum problems, they are far more efficient than SVRG, a popular first-order method.

We present a generic framework for trading off fidelity and cost in computing stochastic gradients when the costs of acquiring stochastic gradients of different quality are not known a priori. We consider a mini-batch oracle that distributes a limited query budget over a number of stochastic gradients and aggregates them to estimate the true gradient. Since the optimal mini-batch size depends on the unknown cost-fidelity function, we propose an algorithm, {\it EE-Grad}, that sequentially explores the performance of mini-batch oracles and exploits the accumulated knowledge to estimate the one achieving the best performance in terms of cost-efficiency. We provide performance guarantees for EE-Grad with respect to the optimal mini-batch oracle, and illustrate these results in the case of strongly convex objectives. We also provide a simple numerical example that corroborates our theoretical findings.

Support Vector Data Description (SVDD) provides a useful approach to construct a description of multivariate data for single-class classification and outlier detection with various practical applications. Gaussian kernel used in SVDD formulation allows flexible data description defined by observations designated as support vectors. The data boundary of such description is non-spherical and conforms to the geometric features of the data. By varying the Gaussian kernel bandwidth parameter, the SVDD-generated boundary can be made either smoother (more spherical) or tighter/jagged. The former case may lead to under-fitting, whereas the latter may result in overfitting. Peak criterion has been proposed to select an optimal value of the kernel bandwidth to strike the balance between the data boundary smoothness and its ability to capture the general geometric shape of the data. Peak criterion involves training SVDD at various values of the kernel bandwidth parameter. When training datasets are large, the time required to obtain the optimal value of the Gaussian kernel bandwidth parameter according to Peak method can become prohibitively large. This paper proposes an extension of Peak method for the case of large data. The proposed method gives good results when applied to several datasets. Two existing alternative methods of computing the Gaussian kernel bandwidth parameter (Coefficient of Variation and Distance to the Farthest Neighbor) were modified to allow comparison with the proposed method on convergence. Empirical comparison demonstrates the advantage of the proposed method.

Regulators require financial institutions to estimate counterparty default risks from liquid CDS quotes for the valuation and risk management of OTC derivatives. However, the vast majority of counterparties do not have liquid CDS quotes and need proxy CDS rates. Existing methods cannot account for counterparty-specific default risks; we propose to construct proxy CDS rates by associating to illiquid counterparty liquid CDS Proxy based on Machine Learning Techniques. After testing 156 classifiers from 8 most popular classifier families, we found that some classifiers achieve highly satisfactory accuracy rates. Furthermore, we have rank-ordered the performances and investigated performance variations amongst and within the 8 classifier families. This paper is, to the best of our knowledge, the first systematic study of CDS Proxy construction by Machine Learning techniques, and the first systematic classifier comparison study based entirely on financial market data. Its findings both confirm and contrast existing classifier performance literature. Given the typically highly correlated nature of financial data, we investigated the impact of correlation on classifier performance. The techniques used in this paper should be of interest for financial institutions seeking a CDS Proxy method, and can serve for proxy construction for other financial variables. Some directions for future research are indicated.

The idea of cross-validation is to split the data into N subsets, to put one subset aside, to estimate parameters of the model from the remaining N-1 subsets, and to use the retained subset to estimate the error of the model. Such a process is repeated N times - with each of the N subsets being used as the validation set . Then the values of the errors obtained in such N steps are combined to provide the final estimate of the model error. The cross-validation is used in various classification and prediction procedures, such as regression analysis, discriminant analysis, neural networks and classification and regression trees (CART) . The goal is to improve the quality of the decision that is made from the outcome of the study on the basis of statistical methods, and to ensure that maximum information is obtained from scarce experimental data.

While this will serve as an overview of Scott's essay, which you can read for further detail and mathematical insights, we will start by with Fortmann-Roe's verbatim definitions which are central to the piece: Error due to Bias: The error due to bias is taken as the difference between the expected (or average) prediction of our model and the correct value which we are trying to predict. Again, imagine you can repeat the entire model building process multiple times. Fortmann-Roe ends the section on over- and under-fitting by pointing to another of his great essays (Accurately Measuring Model Prediction Error), and then moving on to the highly-agreeable recommendation that "resampling based measures such as cross-validation should be preferred over theoretical measures such as Aikake's Information Criteria." I recommend reading Scott Fortmann-Roe's entire bias-variance tradeoff essay, as well as his piece on measuring model prediction error.

K-nearest neighbors, or KNN, is a supervised learning algorithm for either classification or regression. It's super intuitive and has been applied to many types of problems. To make a personalized offer to one customer, you might employ KNN to find similar customers and base your offer on their purchase behaviors. KNN has also been applied to medical diagnosis and credit scoring. This is a post about the K-nearest neighbors algorithm and Python.

In this guide, we'll take a practical, concise tour through modern machine learning algorithms. While other such lists exist, they don't really explain the practical tradeoffs of each algorithm, which we hope to do here. We'll discuss the advantages and disadvantages of each algorithm based on our experience. Categorizing machine learning algorithms is tricky, and there are several reasonable approaches; they can be grouped into generative/discriminative, parametric/non-parametric, supervised/unsupervised, and so on. However, from our experience, this isn't always the most practical way to group algorithms.

In this tutorial we look at the word2vec model by Mikolov et al. This model is used for learning vector representations of words, called "word embeddings". This tutorial is meant to highlight the interesting, substantive parts of building a word2vec model in TensorFlow. This basic example contains the code needed to download some data, train on it a bit and visualize the result. But first, let's look at why we would want to learn word embeddings in the first place.