2021 Feb     Mar     Apr     May     Jun     Jul    
2020 Jan     May     Jun     Jul     Aug     Sep     Oct     Nov     Dec    
2019 Jan     Feb     Mar     Apr     May     Jun     Jul     Aug     Oct     Nov    
2018 Jan     Feb     Mar     Apr     May     Jun     Jul     Aug     Oct     Nov     Dec    
2017 Jan     Feb     Mar     Apr     May     Jun     Jul     Aug     Oct     Nov     Dec    

Event(s) on March 2014

  • Tuesday, 4th March, 2014

    Title: On Invariants of Colored Links
    Speaker: Prof. Boju Jiang, Peking University, China
    Time/Place: 16:30  -  17:30 (Preceded by Reception at 4:00pm)
    SPH, Shiu Pong Hall, Hong Kong Baptist University
    Abstract: The multi-variable Alexander polynomial is the classical invariant for colored links. Conway (1970) refined it to the Conway Potential Function (CPF) with simple skein relations. But it is unclear whether Conway’s relations are enough to characterize CPF. In the 1990's relations are found for this purpose but they are quite complicated. We will introduce a much simpler skein characterization for CPF.

  • Friday, 7th March, 2014

    Title: Classification with unstructured predictors with an application to sentiment analysis
    Speaker: Dr. Junhui WANG, Department of Mathematics, City University of Hong Kong, Hong Kong
    Time/Place: 11:00  -  12:00
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: Unstructured data refers to information that lacks certain structures and cannot be organized in a predefined fashion. Unstructured data involve heavily on words, texts, graphs, objects or multimedia types of files that are difficult to process and analyze by traditional computational tools and statistical methods. In this talk, I will discuss ordinal classification with unstructured predictors and ordered class categories, where imprecise information concerning strengths between predictors is available for predicting the class labels. We integrate the imprecise predictor relations into linear relational constraints over classification function coefficients, where large margin ordinal classifiers are introduced, subject to quadratically many linear constraints. The proposed methods are implemented via a scalable quadratic programming algorithm based on sparse word representations. The advantage is demonstrated in a variety of simulated experiments as well as one large-scale sentiment analysis example on customer reviews. If time permits, the asymptotic properties will also be discussed, which confirm that utilizing relationships among unstructured predictors can significantly improve prediction accuracy.

  • Monday, 17th March, 2014

    Title: On the Importance (and Perils) of Being Skew-Symmetric
    Speaker: Prof. Arieh Iserles, Department of Applied Mathematics and Theoretical Physics, University of Cambridge, United Kingdom
    Time/Place: 15:30  -  16:30 (Preceded by Reception at 3:00pm)
    RRS905, Sir Run Run Shaw Building, HSH Campus, Hong Kong Baptist University
    Abstract: In this talk we go back to the very basics of numerical analysis of PDEs, stability theory of finite difference schemes for linear evolution equations with variable coefficients. We prove that a universal magic wand renders numerical methods stable: the (first) space derivative should be discretised by a skew-symmetric matrix. The downside, however, is a barrier of 2 on the order of such methods on uniform grids. We derive a general theory coupling grid structure with the availability of skew-symmetric matrices corresponding to high-order methods.

  • Tuesday, 25th March, 2014

    Title: Quasi-instrumental variable based inference for non-sparse high-dimensional transformation models
    Speaker: Mr. Xuehu ZHU, Department of Mathematics, Hong Kong Baptist University , Hong Kong
    Time/Place: 11:00  -  12:00
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: For non-sparse high-dimensional linear transformation models for which variable selection methods often lead to biased working models and thus inconsistent estimation of the parameters thereof, we in this paper investigate a novel bias-correction method by applying a quasi-instrumental variable technology. A two-stage estimation procedure is proposed. {A preliminary estimator based on the LASSO is obtained in the first stage, and then a semi-parametric estimator for the bias-corrected working model is developed at the second stage.} A distinguishing feature is that the final estimator is root-$n$ consistent with its asymptotic distribution derived under mild regularity conditions. The performance of the new method is illustrated through a simulation study.



All years