CHENG Li

Multivariate Regression with Gross Errors on Manifold-valued Data

We consider the topic of multivariate regression on manifold-valued output, that is, for a multivariate observation, its output response lies on a manifold. Moreover, we propose a new regression model to deal with the presence of grossly corrupted manifold-valued responses, a bottleneck issue commonly encountered in practical scenarios. Our model first takes a correction step on the grossly corrupted responses via geodesic curves on the manifold, then performs multivariate linear regression on the corrected data.

type: 
Journal Paper
journal: 
IEEE Transactions on Pattern Analysis and Machine Intelligence, 2018, doi: 10.1109/TPAMI.2017.2776260
Url: 
http://ieeexplore.ieee.org/document/8246552/
Impact Factor: 
8.329

Multiview and Multimodal Pervasive Indoor Localization

Pervasive indoor localization (PIL) aims to locate an indoor mobile-phone user without any infrastructure assistance. Conventional PIL approaches employ a single probe (i.e., target) measurement to localize by identifying its best match out of a fingerprint gallery. However, a single measurement usually captures limited and inadequate location features. More importantly, the reliance on a single measurement bears the inherent risk of being inaccurate and unreliable, due to the fact that the measurement could be noisy and even corrupted.

type: 
Conference Paper/Poster
journal: 
"Multi Media 17, Proceedings of the 2017 ACM on Multimedia Conference, Pg 109-117. Mountain view, California, USA, Oct 23-27 2017, doi: 10.1145/3123266.3123436 "
Url: 
https://dl.acm.org/citation.cfm?id=3123266.3123436

Multiview and Multimodal Pervasive Indoor Localization

Pervasive indoor localization (PIL) aims to locate an indoor mobile-phone user without any infrastructure assistance. Conventional PIL approaches employ a single probe (i.e., target) measurement to localize by identifying its best match out of a fingerprint gallery. However, a single measurement usually captures limited and inadequate location features. More importantly, the reliance on a single measurement bears the inherent risk of being inaccurate and unreliable, due to the fact that the measurement could be noisy and even corrupted.

type: 
Conference Paper/Poster
journal: 
Multi Media 17, Proceedings of the 2017 ACM on Multimedia Conference, Pg 109-117. Mountain view, California, USA, Oct 23-27 2017, doi: 10.1145/3123266.3123436
Url: 
https://dl.acm.org/citation.cfm?id=3123266.3123436
Date of acceptance: 
2017-12-29

Quantitative Localization of a Golgi Protein by Imaging Its Center of Fluorescence Mass

The Golgi complex consists of serially stacked membrane cisternae which can be further categorized into sub-Golgi regions, including the cis-Golgi, medial-Golgi, trans-Golgi and trans-Golgi network. Cellular functions of the Golgi are determined by the characteristic distribution of its resident proteins. The spatial resolution of conventional light microscopy is too low to resolve sub-Golgi structure or cisternae. Thus, the immuno-gold electron microscopy is a method of choice to localize a protein at the sub-Golgi level.

type: 
Journal Paper
journal: 
JoVE Journal of Visualized Experiments, 2017, doi:10.3791/55996
Url: 
https://www.jove.com/video/55996
Date of acceptance: 
2017-12-29

Transduction on Directed Graphs via Absorbing Random Walks

In this paper we consider the problem of graph-based transductive classification, and we are particularly interested in the directed graph scenario which is a natural form for many real world applications.Different from existing research efforts that either only deal with undirected graphs or circumvent directionality by means of symmetrization, we propose a novel random walk approach on directed graphs using absorbing Markov chains, which can be regarded as maximizing the accumulated expected number of visits from the unlabeled transient states.

type: 
Journal Paper
journal: 
IEEE Transactions on Pattern Analysis and Machine Intelligence, Issue 99, 2017, doi:10.1109/TPAMI.2017.2730871
pubmed: 
28809671
Url: 
http://ieeexplore.ieee.org/document/8008851/
Impact Factor: 
8.329

Quantitative 3D analysis of complex single border cell behaviors in coordinated collective cell migration

Understanding the mechanisms of collective cell migration is crucial for cancer metastasis, wound healing and many developmental processes. Imaging a migrating cluster in vivo is feasible, but the quantification of individual cell behaviours remains challenging. We have developed an image analysis toolkit, CCMToolKit, to quantify the Drosophila border cell system. In addition to chaotic motion, previous studies reported that the migrating cells are able to migrate in a highly coordinated pattern.

type: 
Journal Paper
journal: 
Nature Communications, 2017 Apr 4;8:14905. doi: 10.1038/ncomms14905
pubmed: 
28374738
Url: 
https://www.nature.com/articles/ncomms14905
Impact Factor: 
12.124
Date of acceptance: 
2017-02-10

Fusion of Magnetic and Visual Sensors for Indoor Localization: Infrastructure-free and More Effective

Accurate and infrastructure-free indoor positioning can be very useful in a variety of applications. However, most

type: 
Journal Paper
journal: 
IEEE Transactions on Medical Imaging, Issue 99, 2016, doi: 10.1109/TMI.2016.2623357
Impact Factor: 
3.756

Segment 2D and 3D Filaments by Learning Sructured and Contextual Features

We focus on the challenging problem of filamentary structure segmentation in both 2D and 3D images, including retinal vessels and neurons, among others. Despite the increasing amount of efforts in learning based methods to tackle this

type: 
Journal Paper
journal: 
IEEE Transactions on Medical Imaging, Issue 99, 2016, doi: 10.1109/TMI.2016.2623357
Impact Factor: 
2.536

Action Recognition in Still Images With Minimum Annotation Efforts

We focus on the problem of still image-based human action recognition, which essentially involves making prediction by analyzing human poses and their interaction with objects in the scene. Besides image-level action labels (e.g., riding, phoning), during both training and testing stages, existing works usually require additional input of human bounding boxes to facilitate the characterization of the underlying human–object interactions. We argue that this additional input requirement might severely discourage potential applications and is not very necessary.

type: 
Journal Paper
journal: 
IEEE Transactions on Image Processing, Vol. 25, Issue 11, Nov 2016, Pg 5479-5490, doi: 10.1109/TIP.2016.2605305
Url: 
http://ieeexplore.ieee.org/document/7558119/
Impact Factor: 
3.735

Lie-X: Depth Image Based Articulated Object Pose Estimation, Tracking, and Action Recognition on Lie Groups

Pose estimation, tracking, and action recognition of articulated objects from depth images are important and challenging problems, which are normally considered separately. In this paper, a uni ed paradigm based on Lie group theory is proposed, which enables us to collectively address these related problems. Our approach is also applicable to a wide range of articulated objects. Empirically it is evaluated on lab animals including mouse and sh, as well as on human hand. On
these applications, it is shown to deliver competitive results compared to the state-of-the-arts, and non-trivial

type: 
Journal Paper
journal: 
International Journal of Computer Vision, 2016
Impact Factor: 
4.207
Syndicate content