Topic sparsity refers to the observation that individual documents usually focus on several salient topics instead of covering a wide variety of topics, and a real topic adopts a narrow range of terms instead of a wide coverage of the vocabulary. Understanding this topic sparsity is especially important for analyzing user-generated web content and social media, which are featured in the form of extremely short posts and discussions. As topic sparsity of individual documents in online social media increases, so does the difficulty of analyzing the online text sources using traditional methods.In this paper, we propose two novel neural models by providing sparse posterior distributions over topics based on the Gaussian sparsemax construction, enabling efficient training by stochastic backpropagation. We construct an inference network conditioned on the input data and infer the variational distribution with the relaxed Wasserstein (RW) divergence. Unlike existing works based on Gaussian softmax construction and Kullback-Leibler (KL) divergence, our approaches can identify latent topic sparsity with training stability, predictive performance, and topic coherence. Experiments on different genres of large text corpora have demonstrated the effectiveness of our models as they outperform both probabilistic and neural methods.

Background: Developing appropriate computational tools to distill biological insights from large-scale gene expression data has been an important part of systems biology. Considering that gene relationships may change or only exist in a subset of collected samples, biclustering that involves clustering both genes and samples has become increasingly important, especially when the samples are pooled from a wide range of experimental conditions. Methods: In this paper, we introduce a new biclustering algorithm to find subsets of genomic expression features (EFs) (e.g., genes, isoforms, exon inclusion) that show strong "group interactions" under certain subsets of samples. Group interactions are defined by strong partial correlations, or equivalently, conditional dependencies between EFs after removing the influences of a set of other functionally related EFs. Our new biclustering method, named SCCA-BC, extends an existing method for group interaction inference, which is based on sparse canonical correlation analysis (SCCA) coupled with repeated random partitioning of the gene expression data set. Results: SCCA-BC gives sensible results on real data sets and outperforms most existing methods in simulations. Software is available at https://github.com/pimentel/scca-bc. Conclusions: SCCA-BC seems to work in numerous conditions and the results seem promising for future extensions. SCCA-BC has the ability to find different types of bicluster patterns, and it is especially advantageous in identifying a bicluster whose elements share the same progressive and multivariate normal distribution with a dense covariance matrix.Keywords: biclustering; SCCA; gene clusters Author summary: In this paper, we introduce a novel biclustering algorithm to find subsets of genomic expression features (EFs) (e.g., genes, isoforms, exon inclusion) that show strong partial correlations (i.e., conditional dependencies between EFs after removing the influences of other EFs in the same set) under certain subsets of samples. We extend an existing method for inferring such conditional dependencies, which is based on sparse canonical correlation analysis (SCCA) coupled with repeated random partitioning and subsampling of the gene expression data set. We incorporate a binary vector such that it will assist the objective function on deciding exclusion or inclusion of a particular sample to the bicluster. We test our algorithm on both simulation and real datasets, and achieve promising results. In addition, our algorithm is shown to be relatively robust to initialization and small perturbation in hyper-parameters. The algorithm is available at https://github.com/ pimentel/scca-bc.

The inconsistency of open pharmacogenomics datasets produced by different studies limits the usage of such datasets in many tasks, such as biomarker discovery. Investigation of multiple pharmacogenomics datasets confirmed that the pairwise sensitivity data correlation between drugs, or rows, across different studies (drug-wise) is relatively low, while the pairwise sensitivity data correlation between cell-lines, or columns, across different studies (cell-wise) is considerably strong. This common interesting observation across multiple pharmacogenomics datasets suggests the existence of subtle consistency among the different studies (i.e., strong cell-wise correlation). However, significant noises are also shown (i.e., weak drug-wise correlation) and have prevented researchers from comfortably using the data directly. Motivated by this observation, we propose a novel framework for addressing the inconsistency between large-scale pharmacogenomics data sets. Our method can significantly boost the drug-wise correlation and can be easily applied to resummarized and normalized datasets proposed by others. We also investigate our algorithm based on many different criteria to demonstrate that the corrected datasets are not only consistent, but also biologically meaningful. Eventually, we propose to extend our main algorithm into a framework, so that in the future when more datasets become publicly available, our framework can hopefully offer a "ground-truth" guidance for references.

In this paper, we propose new proximal Newtontype methods for convex optimization problems in composite form. The applications include model predictive control (MPC) and embedded MPC. Our new methods are computationally attractive since they do not require evaluating the Hessian at each iteration while keeping fast convergence rate. More specifically, we prove the global convergence is guaranteed and the superlinear convergence is achieved in the vicinity of an optimal solution. We also develop several practical variants by incorporating quasi-Newton and inexact subproblem solving schemes and provide theoretical guarantee for them under certain conditions. Experimental results on real-world datasets demonstrate the effectiveness and efficiency of new methods.

The inconsistency of open pharmacogenomics datasets produced by different studies limits the usage of pharmacogenomics in biomarker discovery. Investigation of multiple pharmacogenomics datasets confirmed that the pairwise sensitivity data correlation between drugs, or rows, across different studies (drug-wise) is relatively low, while the pairwise sensitivity data correlation between cell-lines, or columns, across different studies (cell-wise) is considerably strong. This common interesting observation across multiple pharmacogenomics datasets suggests the existence of subtle consistency among the different studies (i.e., strong cell-wise correlation). However, significant noises are also shown (i.e., weak drug-wise correlation) and have prevented researchers from comfortably using the data directly. Motivated by this observation, we propose a novel framework for addressing the inconsistency between large-scale pharmacogenomics data sets. Our method can significantly boost the drug-wise correlation and can be easily applied to re-summarized and normalized datasets proposed by others. We also investigate our algorithm based on many different criteria to demonstrate that the corrected datasets are not only consistent, but also biologically meaningful. Eventually, we propose to extend our main algorithm into a framework, so that in the future when more data-sets become publicly available, our framework can hopefully offer a "ground-truth" guidance for references.