Display advertising business brings billions dollars income yearly in Alibaba. In cost-per-click (CPC) advertising system, advertisements are ranked by the eCPM (effective cost per mille) which is the product of the bid price and CTR (click-through rate). Hence, the performance of CTR prediction model has a straight impact on the final revenue and plays a key role in the advertising system. Driven by the success of deep learning in image recognition, computer vision and natural language processing, a number of deep learning based methods have been proposed for CTR prediction task [1, 2, 3, 4]. These methods usually first employ embedding layer on the input, mapping original large scale sparse id features to the distributed representations, then add fully connected layers (in other words, multilayer perceptrons, MLPs) to automatically learn the nonlinear relations among features. Compared to traditional commonly used logistic regression model [5, 6]. MLPs can reduce a lot of feature engineering jobs, which is time and manpower consuming in industry applications. MLPs now have become a popular model structure on CTR prediction problem. However, in the fields with rich internet-scale user behavior data, such as online advertising and recommendation system in e-commence industry, these MLPs models often lack of deep understanding and exploiting the specific structures of behavior data, thus leave space for further improvement.

It is known that Boosting can be interpreted as a gradient descent technique to minimize an underlying loss function. Specifically, the underlying loss being minimized by the traditional AdaBoost is the exponential loss, which is proved to be very sensitive to random noise/outliers. Therefore, several Boosting algorithms, e.g., LogitBoost and SavageBoost, have been proposed to improve the robustness of AdaBoost by replacing the exponential loss with some designed robust loss functions. In this work, we present a new way to robustify AdaBoost, i.e., incorporating the robust learning idea of Self-paced Learning (SPL) into Boosting framework. Specifically, we design a new robust Boosting algorithm based on SPL regime, i.e., SPLBoost, which can be easily implemented by slightly modifying off-the-shelf Boosting packages. Extensive experiments and a theoretical characterization are also carried out to illustrate the merits of the proposed SPLBoost.

In this paper, we study a new learning paradigm for Neural Machine Translation (NMT). Instead of maximizing the likelihood of the human translation as in previous works, we minimize the distinction between human translation and the translation given by an NMT model. To achieve this goal, inspired by the recent success of generative adversarial networks (GANs), we employ an adversarial training architecture and name it as Adversarial-NMT. In Adversarial-NMT, the training of the NMT model is assisted by an adversary, which is an elaborately designed Convolutional Neural Network (CNN). The goal of the adversary is to differentiate the translation result generated by the NMT model from that by human. The goal of the NMT model is to produce high quality translations so as to cheat the adversary. A policy gradient method is leveraged to co-train the NMT model and the adversary. Experimental results on English$\rightarrow$French and German$\rightarrow$English translation tasks show that Adversarial-NMT can achieve significantly better translation quality than several strong baselines.

Latent Dirichlet Allocation (LDA) is a topic model widely used in natural language processing and machine learning. Most approaches to training the model rely on iterative algorithms, which makes it difficult to run LDA on big data sets that are best analyzed in parallel and distributed computational environments. Indeed, current approaches to parallel inference either don't converge to the correct posterior or require storage of large dense matrices in memory. We present a novel sampler that overcomes both problems, and we show that this sampler is faster, both empirically and theoretically, than previous Gibbs samplers for LDA. We do so by employing a novel Pólya-Urn-based approximation in the sparse partially collapsed sampler for LDA. We prove that the approximation error vanishes with data size, making our algorithm asymptotically exact, a property of importance for large-scale topic models. In addition, we show, via an explicit example, that - contrary to popular belief in the topic modeling literature - partially collapsed samplers can be more efficient than fully collapsed samplers. We conclude by comparing the performance of our algorithm with that of other approaches on well-known corpora. Keywords: Bayesian inference, Big Data, computational complexity, Gibbs sampling, Latent Dirichlet Allocation, Markov Chain Monte Carlo, natural language processing, parallel and distributed systems, topic models.

We study a fundamental class of regression models called the second order linear model (SLM). The SLM extends the linear model to high order functional space and has attracted considerable research interest recently. Yet how to efficiently learn the SLM under full generality using nonconvex solver still remains an open question due to several fundamental limitations of the conventional gradient descent learning framework. In this study, we try to attack this problem from a gradient-free approach which we call the moment-estimation-sequence (MES) method. We show that the conventional gradient descent heuristic is biased by the skewness of the distribution therefore is no longer the best practice of learning the SLM. Based on the MES framework, we design a nonconvex alternating iteration process to train a $d$-dimension rank-$k$ SLM within $O(kd)$ memory and one-pass of the dataset. The proposed method converges globally and linearly, achieves $\epsilon$ recovery error after retrieving $O[k^{2}d\cdot\mathrm{polylog}(kd/\epsilon)]$ samples. Furthermore, our theoretical analysis reveals that not all SLMs can be learned on every sub-gaussian distribution. When the instances are sampled from a so-called $\tau$-MIP distribution, the SLM can be learned by $O(p/\tau^{2})$ samples where $p$ and $\tau$ are positive constants depending on the skewness and kurtosis of the distribution. For non-MIP distribution, an addition diagonal-free oracle is necessary and sufficient to guarantee the learnability of the SLM. Numerical simulations verify the sharpness of our bounds on the sampling complexity and the linear convergence rate of our algorithm.

We obtain the first polynomial-time algorithm for exact tensor completion that improves over the bound implied by reduction to matrix completion. The algorithm recovers an unknown 3-tensor with $r$ incoherent, orthogonal components in $\mathbb R^n$ from $r\cdot \tilde O(n^{1.5})$ randomly observed entries of the tensor. This bound improves over the previous best one of $r\cdot \tilde O(n^{2})$ by reduction to exact matrix completion. Our bound also matches the best known results for the easier problem of approximate tensor completion (Barak & Moitra, 2015). Our algorithm and analysis extends seminal results for exact matrix completion (Candes & Recht, 2009) to the tensor setting via the sum-of-squares method. The main technical challenge is to show that a small number of randomly chosen monomials are enough to construct a degree-3 polynomial with precisely planted orthogonal global optima over the sphere and that this fact can be certified within the sum-of-squares proof system.

Sparse support vector machine (SVM) is a popular classification technique that can simultaneously learn a small set of the most interpretable features and identify the support vectors. It has achieved great successes in many real-world applications. However, for large-scale problems involving a huge number of samples and extremely high-dimensional features, solving sparse SVMs remains challenging. By noting that sparse SVMs induce sparsities in both feature and sample spaces, we propose a novel approach, which is based on accurate estimations of the primal and dual optima of sparse SVMs, to simultaneously identify the features and samples that are guaranteed to be irrelevant to the outputs. Thus, we can remove the identified inactive samples and features from the training phase, leading to substantial savings in both the memory usage and computational cost without sacrificing accuracy. To the best of our knowledge, the proposed method is the \emph{first} \emph{static} feature and sample reduction method for sparse SVM. Experiments on both synthetic and real datasets (e.g., the kddb dataset with about 20 million samples and 30 million features) demonstrate that our approach significantly outperforms state-of-the-art methods and the speedup gained by our approach can be orders of magnitude.

Hierarchical probabilistic models, such as mixture models, are used for cluster analysis. These models have two types of variables: observable and latent. In cluster analysis, the latent variable is estimated, and it is expected that additional information will improve the accuracy of the estimation of the latent variable. Many proposed learning methods are able to use additional data; these include semi-supervised learning and transfer learning. However, from a statistical point of view, a complex probabilistic model that encompasses both the initial and additional data might be less accurate due to having a higher-dimensional parameter. The present paper presents a theoretical analysis of the accuracy of such a model and clarifies which factor has the greatest effect on its accuracy, the advantages of obtaining additional data, and the disadvantages of increasing the complexity.

Data mining techniques are intended to extract information from large volumes of data (Tan et al., 2006). Data mining includes tasks such as classification, regression, clustering, or anomaly detection, among others. Traditional data mining techniques are typically applied to tabulated data. Novel techniques have also been devised for semi-structured or structured data, since exploiting the relationships among instances from a dataset leads to new research and development opportunities (Getoor and Diehl, 2005). For example, network data mining has been used to predict previously unknown protein interactions in protein-protein interaction networks (Martínez et al., 2014). It has also been used to study and predict future author collaborations and tendencies in co-authorship networks (Pavlov and Ichise, 2007). Different network mining techniques are used by popular internet search engines to rank the most relevant websites (Page et al., 1999). These are only some examples of the large number of applications of network data mining. There are many software tools that facilitate the analysis of networked data.

Santiago Ramón y Cajal, a Spanish histologist and anatomist known today as the father of modern neuroscience, was also a committed psychologist who believed psychoanalysis and Freudian dream theory were "collective lies." When Freud published The Interpretation of Dreams in 1900, the science world swooned over his theory of the unconscious. Dreams quickly became synonymous with repressed desire. Puzzling dream images could unlock buried conflicts, the psychoanalyst said, given the correct interpretation. Cajal, who won the 1906 Nobel Prize for discovering neurons and, more remarkably, intuiting the form and function of synapses, set out to prove Freud wrong. To disprove the theory that every dream is the result of a repressed desire, Cajal began keeping a dream journal and collecting the dreams of others, analyzing them with logic and rigor. Translated here into English for the first time, the dreams of Santiago Ramón y Cajal offer insight into the mind of a great scientist. Cajal eventually deemed the project unpublishable.