[image 01806] Talk by Guo-Jun Qi at 1509 in NII (11:00-12:00, on May 19 Thursday)

[Yi Yu]

Dear All,

The following talk will be delivered on May 19 (Thursday) in NII.

Best Regards,
Yi Yu

---------------------------------------------------------------------------
Date: May 19 (Thursday), 2016
Time: 11:00-12:00
Place: Room 1509 in NII (http://www.nii.ac.jp/en/about/access/)
Speaker: Guo-Jun Qi

Title: Hierarchically Gated Deep Networks for Semantic Segmentation

Abstract: Semantic segmentation aims to parse the scene structure of images
by annotating the labels to each pixel so that images can be segmented into
different regions.   While image structures usually have various scales, it
is difficult to use a single scale to model the spatial contexts for all
individual pixels.  Multi-scale Convolutional Neural Networks (CNNs) and
their variants have made striking success for modeling the global scene
structure for an image. However, they are limited in labeling fine-grained
local structures like pixels and patches, since spatial contexts might be 
blindly mixed up without appropriately customizing their scales. To
address this challenge, we develop a novel paradigm of multi-scale deep
network to model spatial contexts surrounding different pixels at various
scales.  It builds multiple layers of memory cells, learning feature
representations for individual pixels at their customized scales by
hierarchically absorbing relevant spatial contexts via memory gates between
layers.
Such Hierarchically Gated Deep Networks (HGDNs) can customize a suitable
scale for each pixel, thereby delivering better performance on labeling
scene structures of various scales.  We conduct the experiments on two
datasets, and show competitive results compared with the other multi-scale
deep networks on the semantic segmentation task.


Bio:
Dr. Guo-Jun Qi is an assistant professor in the Department of Computer Science at
the University of Central Florida.  His research interests include
knowledge discovery, analysis and aggregation of big data deluging from a
variety of modalities and sources in order to build smart and reliable
information and decision-making systems.  He strives to apply my research
to solve the practical problems through high quality data processing and
analysis in healthcare, sensor and social networks, financial systems and
so forth. He was the recipient of one-time Microsoft Fellowship, and twice
IBM Fellowships.

Dr. Qi has published over 60 papers in a wide range of venues, such as
Proceedings of IEEE, IEEE T PAMI, IEEE T KDE, IEEE T Image Processing, ACM
SIGKDD, WWW, ICML, ACM MM, CVPR, ICDM, SDM and ICDE.  Among them are the
best paper of ICDM 2014 Best Student Paper,“the best ICDE 2013 paper” by
IEEE Transactions on Knowledge and Data Engineering, as well as ACM
Multimedia 2007. His publications have been cited 2500+ times and his
current h-index is 24 (by Google Scholar).

He has served or will serve as an area chair (a senior program committee
member) for ACM SIGKDD Conference on Knowledge Discovery and Data Mining
(KDD), ACM International Conference on Information and Knowledge Management
(CIKM), as well as International ACM Conference on Multimedia (ACM MM). He
is also serving or have served in the program committees of several
academic conferences, including CVPR, ICCV, KDD, WSDM, CIKM, IJCAI, ICMR,
ACM Multimedia, ACM/IEEE ASONAM, ICDM, ICIP, and ACL. He is a guest/lead
editor for the special issue on Big Media Data: Understanding, Search, and
Mining" in IEEE Transactions on Big Data, “Deep Learning for Multimedia
Computing” in IEEE Transactions on Multimedia, and the special issue on
“Social Media Mining and Knowledge Discovery” of Multimedia Systems,
Springer. In addition, he was a Program Co-Chair for International
Conference on MultiMedia Modeling (Springer) 2016, a Co-chair for
International ACM Workshop on Crowdsourcing for Multimedia (CrowdMM) 2015
and IEEE International Workshop on Frontier of Crowdsourcing in Multimedia
Computing (FCMC) 2014.