When: Wednesday, February 6, 2019
Where: PGH 563
Time: 11:00 AM

Study and Mitigation of Non-Cooperative UWB Interference on Ranging

Speaker: Hessam Mohammadmoradi

UWB localization systems are well known for their accuracy for indoor localization. Recently UWB-based localization systems with less than 5 cm error have been proposed. Since IEEE802.15.4-11 standard only allows UWB signal transmission with very low power (-41.3 dBm/MHz), UWB systems cannot utilize carrier sensing techniques to manage shared access to the wireless channel. In this paper, first, we studied the likelihood of wideband interference in localization applications and also analyzed the impact of interference on ranging performance. Then, we design and evaluate RAPSI (Random Pulse Shape Identification), a simple yet effective technique to detect and mitigate the ranging error caused by interference. Our results show 30% to 40% reduction in ranging errors caused by wideband interference after applying the proposed techniques.

Bio:

Hessam Mohammadmoradi, is a PhD student at department of computer science at University of Houston working under supervision of Prof. Omprakash Gnawali. His research interests include distributed systems, wireless networks, and Internet of Things. His PhD thesis is focused on robust UWB-based indoor localization systems.

Open Source Face Recognition Performance Evaluation Package

Speaker: Xiang Xu

Biometrics-related research has been accelerated significantly by deep learning technology. However, there are limited open-source resources to help researchers evaluate their deep learning-based biometrics algorithms efficiently, especially for the face recognition tasks. In this work, we design and implement a light-weight, maintainable, scalable, general- izable, and extendable face recognition evaluation toolbox named FaRE that supports both online and offline evaluation to provide feedback to algorithm development and accelerate biometrics-related research. FaRE consists of a set of evaluation metric functions and provides various APIs for commonly-used face recognition datasets including LFW, CFP, UHDB31, and IJB-series datasets, which can be easily extended to include other customized datasets.

Bio:

Xiang Xu is a Ph.D. candidate in Computational Biomedicine Lab (CBL), Department of Computer Science, University of Houston. Before arriving at UH, he received B.Eng. in Beijing University of Posts and Telecommunications. His research mainly focuses on Computer Vision, Deep Learning, Machine Learning, Biometrics, and Face Recognition. He have worked on several projects covering a wide range of topics such as Face Detection, Face Alignment, 3D Face Reconstruction, and Face Recognition. In CBL, he leads the development of a deep learning based 3D-aided 2D face recognition system. Additionally, he have two summer internship experiences in Amazon AI and will join Amazon Rekognition team in July, 2019.

Imbalance Data Classification in Financial Systems:  Challenges and Approaches

Speaker: Hadi Mansourifar

The imbalanced class distribution naturally exists in many financial datasets, where normal instances dramatically outnumber those with abnormal condition. A common technique used to combat consequent unsatisfactory results is to increase the size of a dataset through data augmentation. GANs have been used for data augmentation in several medical image classification problems. An uninvestigated domain where GANs could be highly effective for data augmentation is financial data classification. We took the first step towards GAN-based data augmentation for increasing bankruptcy classification performance successfully. However, GAN-based data augmentation cannot guarantee the best efficiency results in other financial classification problems like credit card fraud detection and default probability prediction. The main reasons are low dimensionality and lack of sufficient minority instances. To address this problem, we propose a method called Deep Synthetic Minority Oversampling Technique which is a novel variation of traditional SMOTE with initial promising experimental results

Bio:

Hadi Mansourifar is a third year PhD student. His advisor is Dr. Shi and his current research is imbalance data classification in financial systems.