Which machine learning algorithm should I use?

Recitations from Tel-Aviv University introductory course to computer science, assembled as IPython notebooks by Yoav Ram. Exploratory Computing with Python, a set of 15 Notebooks that cover exploratory computing, data analysis, and visualization. No prior programming knowledge required. Each Notebook includes a number of exercises (with answers) that should take less than 4 hours to complete. Developed by Mark Bakker for undergraduate engineering students at the Delft University of Technology.

I finally beat the S&P 500 by 10%. This might not sound like much but when we're dealing with large amounts of capital and with good liquidity, the profits are pretty sweet for a hedge fund. More aggressive approaches have resulted in much higher returns. It all started after I read a paper by Gur Huberman titled "Contagious Speculation and a Cure for Cancer: A Non-Event that Made Stock Prices Soar," (with Tomer Regev, Journal of Finance, February 2001, Vol. "A Sunday New York Times article on a potential development of new cancer-curing drugs caused EntreMed's stock price to rise from 12.063 at the Friday close, to open at 85 and close near 52 on Monday. It closed above 30 in the three following weeks. The enthusiasm spilled over to other biotechnology stocks. The potential breakthrough in cancer research already had been reported, however, in the journal Nature, and in various popular newspapers including the Times! Thus, enthusiastic public attention induced a permanent rise in share prices, even though no genuinely new information had been presented."

Diffusion maps are a nonlinear manifold learning technique based on harmonic analysis of a diffusion process over the data. Out-of-sample extensions with computational complexity $\mathcal{O}(N)$, where $N$ is the number of points comprising the manifold, frustrate applications to online learning applications requiring rapid embedding of high-dimensional data streams. We propose landmark diffusion maps (L-dMaps) to reduce the complexity to $\mathcal{O}(M)$, where $M \ll N$ is the number of landmark points selected using pruned spanning trees or k-medoids. Offering $(N/M)$ speedups in out-of-sample extension, L-dMaps enables the application of diffusion maps to high-volume and/or high-velocity streaming data. We illustrate our approach on three datasets: the Swiss roll, molecular simulations of a C$_{24}$H$_{50}$ polymer chain, and biomolecular simulations of alanine dipeptide. We demonstrate up to 50-fold speedups in out-of-sample extension for the molecular systems with less than 4% errors in manifold reconstruction fidelity relative to calculations over the full dataset.

Machine Learning (ML) has been successfully applied to a wide range of domains and applications. One of the techniques behind most of these successful applications is Ensemble Learning (EL), the field of ML that gave birth to methods such as Random Forests or Boosting. The complexity of applying these techniques together with the market scarcity on ML experts, has created the need for systems that enable a fast and easy drop-in replacement for ML libraries. Automated machine learning (autoML) is the field of ML that attempts to answers these needs. Typically, these systems rely on optimization techniques such as bayesian optimization to lead the search for the best model. Our approach differs from these systems by making use of the most recent advances on metalearning and a learning to rank approach to learn from metadata. We propose autoBagging, an autoML system that automatically ranks 63 bagging workflows by exploiting past performance and dataset characterization. Results on 140 classification datasets from the OpenML platform show that autoBagging can yield better performance than the Average Rank method and achieve results that are not statistically different from an ideal model that systematically selects the best workflow for each dataset. For the purpose of reproducibility and generalizability, autoBagging is publicly available as an R package on CRAN.

In the setting of high-dimensional linear regression models, we propose two frameworks for constructing pointwise and group confidence sets for penalized estimators which incorporate prior knowledge about the organization of the non-zero coefficients. This is done by desparsifying the estimator as in van de Geer et al. [18] and van de Geer and Stucky [17], then using an appropriate estimator for the precision matrix $\Theta$. In order to estimate the precision matrix a corresponding structured matrix norm penalty has to be introduced. After normalization the result is an asymptotic pivot. The asymptotic behavior is studied and simulations are added to study the differences between the two schemes.

Mini-batch stochastic gradient descent and variants thereof have become standard for large-scale empirical risk minimization like the training of neural networks. These methods are usually used with a constant batch size chosen by simple empirical inspection. The batch size significantly influences the behavior of the stochastic optimization algorithm, though, since it determines the variance of the gradient estimates. This variance also changes over the optimization process; when using a constant batch size, stability and convergence is thus often enforced by means of a (manually tuned) decreasing learning rate schedule. We propose a practical method for dynamic batch size adaptation. It estimates the variance of the stochastic gradients and adapts the batch size to decrease the variance proportionally to the value of the objective function, removing the need for the aforementioned learning rate decrease. In contrast to recent related work, our algorithm couples the batch size to the learning rate, directly reflecting the known relationship between the two. On popular image classification benchmarks, our batch size adaptation yields faster optimization convergence, while simultaneously simplifying learning rate tuning. A TensorFlow implementation is available.

EARNING a representation for nodes in a graph, also known as node embedding, has been an important tool for extracting features that can be used in machine learning problems involving graph-structured data [1]-[4]. Perhaps the most widely adopted node embedding is the one based on the eigendecomposition of the adjacency matrix or the graph Laplacian [2], [5], [6]. Recent advances in word embeddings for natural language processing such as [7] has inspired the development of analogous embeddings for nodes in graphs [3], [8]. These so-called "neural" node embeddings have been applied to a number of supervised learning problems such us link prediction and node classification and demonstrated stateof-the-art performance [3], [4], [8]. In contrast to applications to supervised learning problems in graphs, in this work we leverage the neural embedding framework to develop an algorithm for the unsupervised community discovery problem in graphs [9]-[12]. The key idea is straightforward: learn node embeddings such that vectors of similar nodes are close to each other in the latent embedding space. Then, the problem of discovering communities in a graph can be solved by finding clusters in the embedding space. We focus on non-overlapping communities and validate the performance of the new approach through a comprehensive set of experiments on both synthetic and real-world data.

Estimating the log-likelihood gradient with respect to the parameters of a Restricted Boltzmann Machine (RBM) typically requires sampling using Markov Chain Monte Carlo (MCMC) techniques. To save computation time, the Markov chains are only run for a small number of steps, which leads to a biased estimate. This bias can cause RBM training algorithms such as Contrastive Divergence (CD) learning to deteriorate. We adopt the idea behind Population Monte Carlo (PMC) methods to devise a new RBM training algorithm termed Population-Contrastive-Divergence (pop-CD). Compared to CD, it leads to a consistent estimate and may have a significantly lower bias. Its computational overhead is negligible compared to CD. However, the variance of the gradient estimate increases. We experimentally show that pop-CD can significantly outperform CD. In many cases, we observed a smaller bias and achieved higher log-likelihood values. However, when the RBM distribution has many hidden neurons, the consistent estimate of pop-CD may still have a considerable bias and the variance of the gradient estimate requires a smaller learning rate. Thus, despite its superior theoretical properties, it is not advisable to use pop-CD in its current form on large problems.

In the first part of this series, I introduced the Outbrain Click Prediction machine learning competition. That post described some preliminary and important data science tasks like exploratory data analysis and feature engineering performed for the competition, using a Spark cluster deployed on Google Dataproc. In this post, I describe the competition evaluation, the design of my cross-validation strategy and my baseline models using statistics and trees ensembles. In that competition, Kagglers were required to rank recommended ads by decreasing predicted likelihood of being clicked. Sponsored search advertising, contextual advertising, display advertising and real-time bidding auctions have all relied heavily on the ability of learned models to predict ad click–through rates (CTRs) accurately, quickly and reliably.