Monthly Archives: September 2015

The segmentation research in our lab made a significant breakthrough these days. The paper “Robust Image Segmentation Using Contour-guided Color Palettes” by Xiang Fu, Chien-Yi Wang, Chen Chen, Changhu Wang and C.-C. Jay Kuo was accepted by ICCV 2015. In this paper, the contour-guided color palette (CCP) is proposed to efficiently integrate contour and color cues of an image. To find representative colors of an image, color samples along long contours between regions, similar in spirit to machine learning methodology that focus on samples near decision boundaries, are collected to achieve an image-dependent color palette. This color palette provides a preliminary segmentation in the spatial domain, which is further fine-tuned by post-processing techniques such as leakage avoidance, fake boundary removal, and small region mergence. While CCP offers an acceptable standalone segmentation result, it can be further integrated into the framework of layered spectral segmentation to produce a more robust segmentation.

For more details, please refer to our paper that will be published soon. The latest code of this paper can be downloaded here. It is written in MATLAB, and has been tested under 64-bit Windows, Linux, and Mac OSX.

Congratulations to Sanjay Purushotham and Pang-Chang (Brian) Lan, who passed their defense last week! Following are their thesis abstracts and they also shared their PhD experiences.
Thesis: Advanced Machine Learning Techniques for Video, Social, and Biomedical Data Analytics (Sanjay Purushotham)
In this thesis, advanced machine learning techniques are developed to tackle challenging problems arising in three Big Data application domains. They are: 1) partial near-duplicate content copy detection and alignment for the multimedia (video) application, 2) personalized single user and group recommender systems for the social media data application, and 3) sparse learning models for identification of discriminative feature interactions for gene expression prediction and cancer stage classification for the biomedical data application. Novel and suitable machine learning algorithms and models are designed to meet the nature of the data in each specific application domain. 

Thesis: Secure Wireless Communications with Side Information: Secrecy Analysis and Unitary Modulation Realization (Brian Lan)
In wireless communication systems such as LTE, information security is protected by symmetric cryptography which assumes that the user and the base station shares the same secret key. But before the setup of the symmetric cryptosystems, the key sharing and authentication processes are vulnerable to eavesdropping. This is referred to as the secure initiation problem. Our work proposes to use physical layer techniques to achieve perfect secrecy without using cryptography, so the secure initiation problem can be solved. In particular, we bring up the concept of having only channel state information (CSI) at the transmitter side but not at the receiver nor the eavesdropper. This concept will be shown beneficial in secrecy comparing to the conventional assumption of full CSI at all terminals. A practical scheme, unitary modulation, is then proposed to exploit this concept. [...]