Pandong Yang
864-***-**** adkdni@r.postjobfree.com 3306 Thompson Farm, Bedford, MA, 01730
LinkedIn: https://linkedin.com/in/pandong-yang-babbb41b9/ GitHub: https://github.com/pandongy
Education
Tufts University Aug. 2019 – May 2021
M.S., Data Science GPA: 3.92/4.00 Medford, MA
Relative Courses: Deep Neural Network, Intro to Machine Learning, Nature Language Processing, Artificial In- telligence, Probabilistic System Analysis, Big Data, Principles Data Science in Python, Database System University of Southern California Aug. 2010 – May 2012 M.S., Industrial and Systems Engineering GPA 3.63/4.00 Los Angeles, CA Relative Courses: ERP System and Application, Financial Engineering, Engineering Project Management, Per- formance Analysis, Production and Scheduling, Designing Spreadsheet-Based Business Models Fudan University Aug. 2006 – July 2010
B.S., Mechanics GPA 3.65/4.00 Shanghai, China
Technical Skills
Professional Emphases: Data Analysis, Data Visualization, Machine Learning, Deep Learning, Machine Learning Models, Nature Language Processing, Image Processing Programming & Libraries: Python, Scala, SQL, R, SKLearn, Keras, Tensorflow, CSBDeep, Pandas, NumPy, Matplotlib Platforms & Software: Google Cloud Platform, Google Colab, Hadoop, Tableau, Thomson Data Analyzer, Excel Languages: Chinese/Mandarin, English
Employment
National Science Library June 2012 – Aug. 2014
Data Analyst Beijing, China
• Academic literature and patent data mining and statistic analysis.
• Generated reports on the latest subject trends and global research hot-spots.
• Developed training courses for researchers on data acquisition, management, and analysis tools.
Key Projects Aug. 2019 – Present
Biomedical Image Denoising
• Cooperated with Prof. Georgakoudi’s optical diagnostics research lab to denoise diseased tissue images by exploiting deep learning techniques.
• Utilized both ground truth and Noise2Noise image denoising methods to train multiple denoising models based on a residual U-net architecture.
• Worked closely with biomedical researchers to continuously optimize models by evaluating the denoised images. Query System Towards COVID-19 Researches
• Employed the Nature Language Processing and Machine Learning techniques (ELMo, k-nearest neighbors) to develop a query system to retrieve related academic articles from the COVID–19 Open Research Data Set.
• Embedded BERT Summarizer in the query system to summarize the retrieved articles. Comparison of Various Classification Methods for Sentiment Reviews
• Explored the impact of different preprocessing methods on sentiment classification results.
• Built three models (logistic regression, multilayer perceptron, and random forest) with well-tuned hyper-parameters to do sentiment classification and analyzed the results. Exploring Different Action-value Methods’ Performance for Non-stationary Problem
• Combined the incremental sample average method with three different action selection methods (greedy method, ϵ-greedy method, and greedy optimal initial method) to solve the non-stationary 10-arm testbed problem.
• Switched to the exponential recency-weighted average method to combine with the same action selection methods used in the first part to solving the same non-stationary 10-arm testbed problem.
• Analyzed and compared the above two action-value methods’ performance.