Colloquium/Seminar

YearMonth
2021 Feb     Mar     Apr     May     Jun     Jul     Aug    
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 October 2019


  • Thursday, 3rd October, 2019

    Title: Simultaneous Registration and Modelling for Multi-dimensional Functional Data
    Speaker: Dr Jian Qing Shi, School of Mathematics, Statistics & Physics, University of Newcastle, United Kingdom
    Time/Place: 09:00  -  10:00
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: The clustering and classification for functional data with misaligned problems has drawn much attention in the last decade. Most methods do the clustering/classification after those functional data being registered and there has been little research using both functional and scalar variables. In this talk, I will discuss a new approach allowing the use of both types of variables and also allowing simultaneous registration and clustering/classification. Numerical results based on both simulated data and real data will be presented.


  • Thursday, 3rd October, 2019

    Title: A Class of Smooth Exact Penalty Function Methods for Optimization Problems with Orthogonality Constraints
    Speaker: Prof. LIU Xin, Institute of Computational Mathematics, Chinese Academy of Sciences, China
    Time/Place: 11:00  -  12:00
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: Updating the augmented Lagrangian multiplier by closed-form expression yields efficient first-order infeasible approach for optimization problems with orthogonality constraints. Hence, parallelization becomes tractable in solving this type of problems. Inspired by this closed-form updating scheme, we propose an exact penalty function model with compact convex constraints (PenC). We show its equivalence to optimization problems with orthogonality constraints under mild condition. Based on PenC, we first propose a first-order algorithm called PenCF and establish its global convergence and local linear convergence rate under some mild assumptions. If the computation and storage of Hessian is achievable, and we pursue high precision solution and fast local convergence rate, a second-order approach called PenCS is proposed under the same penalty function. To avoid expensive calculation or solving a hard subproblem in computing the Newton step, we propose a new strategy to do it approximately which leads to quadratic convergence theoretically. Moreover, the main iterations of both PenCF and PenCS are orthonormalization-free and hence parallelizable. Numerical experiments illustrate that PenCF is comparable with existing first-order methods including the existent infeasible approaches. Furthermore, PenCS shows its stability and high efficiency in obtain high precision solution in comparing with the existent second-order methods.


  • Thursday, 10th October, 2019

    Title: Mixed Noise Removal by Integrating Variational and Deep Learning Methods
    Speaker: Prof Liu Jun, Department of Mathematics, Beijing Normal University, Beijing, China
    Time/Place: 16:00  -  17:00
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: In this talk, the traditional model based variational methods and deep learning based algorithms are integrated to address mixed noise removal, especially for the additive mixture and additive & multiplicative mixture noise removal problems. To be different from single type noise (e.g. Gaussian) removal, it is a challenge problem to accurately discriminate noise types and levels for each pixel. We propose a variational method to iteratively estimate the noise parameters, and then the algorithm can automatically classify the noise according to the different statistical parameters. To enforce the regularization, the deep learning method is employed to learn the natural images prior. Compared with some model based regularizations, the deep convolution neural network (DCNN) can significantly improve the quality of the restored images. Compared with some DCNN based methods, the synthesis step can produce better reconstructions by analyzing the types and levels of the recognized noise. Numerical experiments show that our method can achieve some state-of-the-art results for mixture noise removal.


  • Monday, 21st October, 2019

    Title: Theory and application of inverse source and inverse scattering problem: Geometric body generation and imaging by interior resonant modes
    Speaker: Ms HE Youzhi (何酉子), Department of Mathematics, Hong Kong Baptist University, HKSAR
    Time/Place: 14:30  -  16:00
    FSC1217
    Abstract: Both inverse source and inverse scattering problem play a central role in many scientific fields with important medical, industrial and military applications. I mainly concern about the following two topics: (1) geometric body generation based on partial input parameters by the time-harmonic or time-dependent inverse source problem; (2) reconstruction of the shape of the scatterer with different physical properties by interior resonant modes. I consider both theoretical and computational perspective of these problems.


  • Wednesday, 30th October, 2019

    Title: End-to-end learning for sign language recognition
    Speaker: Mr. ZHOU Mingjie (周明杰), Department of Mathematics, Hong Kong Baptist University, HKSAR
    Time/Place: 15:00  -  16:30
    FSC1217, Fong Shu Chuen Library, HSH Campus, Hong Kong Baptist University
    Abstract: Sign language recognition is useful for bridging the communication gap between sign language users and normal hearing people. This work presents end-to-end frameworks for weakly supervised continuous sign language recognition. According to recent public data sets, these videos are not frame-by-frame labelled and these labels are noisy, which makes recognition difficult. This work will go through current literature and exploit the autoencoder, RNNs and attention-mechanism based sequence models for sign language recognition.

 

 


All years