Xiangyi Meng

Student, Programmer, Primary Researcher

Overview

I am currently a undergraduate student in Department of Computer Science at Xiamen University since 2014. My hometown is Shijiazhuang, Hebei, China. My research mainly focuses on Data Mining, Bioinformatics and Security. My current interests are in graph mining, human action recognition and semi-supervised learning. And I am preparing to study in Hong Kong after graduating from XMU.
The full pdf version of my resume can be downloaded here.

Education

2014-2018 Xiamen University, Xiamen, China, B.E. in Computer Science

Publications

2017

1. Xiangyi Meng, Rui Xu, Xuantong Chen, *Lingxiang Zheng, et al., Human Action Classification in Basketball: A Single Inertial Sensor Based Framework, The 6th International Conference on Frontier Computing (FC 2017), Osaka, Japan
2. Yizhen Wang, Xiangyi Meng, *Lingxiang Zheng, et al., A Smartphone Inertial Sensor Based Recursive Zero-Velocity Detection Approach, The 6th International Conference on Frontier Computing (FC 2017), Osaka, Japan

Selected Awards

2014-2015 National Scholarship
2014-2015 Outstanding Merit Student, Xiamen University
2015-2016 Excellent Young Volunteer, Xiamen University
2016.1 Successful Participant, 2016 Interdisciplinary Contest In Modeling
2016.8 First prize of the 6th “Huawei Cup” AI Designing Competition of University Students in China
2016-2017 Excellent Student Cadre, Xiamen University
2016-2017 Scholarship for Academic Innovation, Xiamen University
2016-2017 Xiamen International Bank Scholarship, Xiamen University

Researches

A keyword-based auto generating system for Chinese couplets

Advisor: Prof. Xuling Zheng

• Novelly expand the current method for couplets generating based on a given sentence to an approach using given keyword.
• A phrase-based SMT model is deployed to generate the initial couplet pairs and an approach for optimizing the quality of the couplets based on immune algorithm is proposed.
• An SVM is trained for the evaluation of the couplets using manually pre-labeled couplets.

A basketball shooting training and motion tracking system via machine learning

Advisor: Lingxiang Zheng, Senior Engineer

• A neural network embedded on Intel Curie is applied for the evaluation of the shooting performance using features extracted from accelerometer and gyrometer data.
• Inspired by Robotics, we constrained the possible location of the joints in the arm into certain spaces represented as point-clouds, aiming at simulating the shooting motion.
• A modified hidden Markov Model is proposed to track the motion of shooting, which its states are defined in the spaces of the point-clouds.

Iteratively collective prediction of disease-gene associations through the incomplete network

Advisor: Prof. Xiangxiang Zeng

• We studied the similarity measure of the nodes in heterogeneous network and came up with “weighted path count with random walk” to fit the complex topological architecture.
• PU learning is studied and integrated into the classification method in view of negative samples in gene-disease link datasets are extremely scarce.
• Being different with state-of-the-art singleton approach, an iterative framework for link prediction on heterogeneous network is proposed and deployed on gene-disease link prediction.
• The source code is published at my GitHub:xymeng16/ISL4LP.

A Django based online Q&A system

Xiamen University Turing Class Online System

• Inspired by Stackoverflow.com, we developed an similar online QA website based on the Django framework to provide a platform for the teachers and students at XMU to communicate with each other.
• Currently, our project is still being developing, and it has 4 basic functions: User Control, Q&A, Online Test, File Upload&Download.
• Our project is deployed on a Ubuntu server at Ali Cloud using uWSGI and Nginx.

Skills

Experienced: C, C++, C#, MATLAB, Latex, Python, Git, Linux
Amateur: Java, Web Development, Tensorflow

Last Updated: May 2017