Workshop on Applications of Deep Learning | April 16th, 2022
Abstract
A class of short-term recurrence Anderson mixing schemes and their applications
Bao, Chenglong, Tsinghua University
包承龙，清华大学
Abstract: In this talk, I will discuss our recent progress on developing a modified Anderson mixing schemes with short memory requirement for solving linear, nonlinear systems and stochastic programming. The convergence analysis will be reported and extensive numerical results will be present.
Supervised and unsupervised learning methods for CT image reconstruction
Ding, Qiaoqiao, Shanghai Jiao Tong University
丁乔乔，上海交通大学
Abstract: Image reconstruction from down-sampled and corrupted measurements, such as fast MRI and low dose CT, is mathematically ill-posed inverse problem. Deep neural network (DNN) has been becoming a prominent tool in the recent development of medical image reconstruction methods. In this talk, I will introduce two works on incorporating classical image reconstruction method and deep learning method. In the first work, we proposed a multi-scale DNN for sparse view CT reconstruction, which directly learns an interpolation scheme to predict the complete set of 2D Fourier coefficients in Cartesian coordinates from the given measurements in polar coordinates. In the second work, we proposed an unsupervised deep learning method for LDCT image reconstruction, which does not require any external training data. The proposed method is built on a re-parameterization technique for Bayesian inference via deep network with random weights, combined with additional total variational (TV) regularization. The experiments on both sparse CT and low dose CT problem show that the proposed method provided state-of-the-art performance.
Unsupervised learning driven by Langevin dynamics and its applications
Li, Ji , National University of Singapore
李季，新加坡国立大学
Abstract: From the Bayesian view, the key component of image restoration is to estimate the posterior distribution. Generally, the sampling from posterior distribution is intractable. To this end, there have been some variational approaches to approximate the posterior distribution using a proxy distribution. In this talk, we first review the Langevin dynamics as an effective sampler for a given distribution. Then we apply it or embed it to the unsupervised learning solution to two image restoration problems with slight modifications.
Image reconstruction using generative models
Liu, Jiulong, Academy of Mathematics and Systems Science,CAS
刘九龙，中国科学院数学与系统科学研究院
Abstract: As the advances in imaging modalities in which the image reconstruction problems are mathematically inverse problems, many new kinds of inverse problems have emerged and trends to being with low-cost data acquisition. In order to efficiently and stably solve the under-determined and ill-conditioned inverse problems with fewer measurements, we established some image reconstruction methods using generative priors which are shown much more efficient than the traditional priors or some other data-driven priors. In this talk, I will introduce some of these methods and present our recent results for MRI reconstruction, phase retrieval, and some other nonlinear inverse problems.
Unsupervised deep learning for image restoration
Pang, Tongyao, National University of Singapore
庞彤瑶，新加坡国立大学
Abstract: Recently, deep learning has become a prominent tool for solving image restoration problems. Most of the deep learning methods are supervised, which requires the collection of paired truth/measurement data. In some scenarios, it is challenging and even impossible to collect such data pairs containing the clean images. To relax the requirement on data collection, we proposed several unsupervised deep learning methods that only use the obtained measurements for training. Our methods are wide applicable to many image restoration problems and achieve state of the art performance..
Deep Gaussian kernel mixture learning for single image defocus deblurring
Quan,Yuhui, South China University of Technology
全宇晖，华南理工大学
Abstract: Defocus blur is one kind of blur effects often seen in images, which is challenging to remove due to its spatially variant amount. This talk presents an end-to-end deep learning approach for removing defocus blur from a single image, so as to have an all-in-focus image for consequent vision tasks. First, a pixel-wise Gaussian kernel mixture (GKM) model is proposed for representing spatially variant defocus blur kernels in an efficient linear parametric form, with higher accuracy than existing models. Then, a deep neural network called GKMNet is developed by unrolling a fixed-point iteration of the GKM-based deblurring. The GKMNet is built on a lightweight scale-recurrent architecture, with a scale-recurrent attention module for estimating the mixing coefficients in GKM for defocus deblurring. Extensive experiments show that the GKMNet not only noticeably outperforms existing defocus deblurring methods, but also has its advantages in terms of model complexity and computational efficiency.
TBD
Xiang, Xueshuang, Qian Xuesen Laboratory, CAST
向雪霜，中国空间技术研究院钱学森实验室
Abstract: TBD.
