Ximin Lin

xlin32@illinois.edu

Junior in Computer Engineering in UIUC

 

Overview

I am a Junior major in computer engineering in UIUC. I am interested in many applications of machine learning.

I am actively looking for internship

 

projects

Android App for Object Recognition, Fall 2017

1.     Implement a SIFT (scale-invariant feature transform) algorithm in Android studio and can do simple object recognitions

2.     Final Group Project for Course Digital Signal Processing Laboratory

3.     Algorithm based on the paper "Distinctive Image Features from Scale-Invariant Keypoints" (International Journal of Computer Vision, 2004.)

4.     Demo Link: https://youtu.be/AShsPyn6eu4

 

 

 

 

FPGA game "I wanna 385", Fall 2017

1.     FPGA (field-programmable gate array)

2.     Final Group Project for Course Digital System Laboratory

3.     Demo Link: https://youtu.be/K5iLkwr95oA

 

 

 

 

research experience

ID Attack: Attack A Chinese Mobile App that can recognize ID with high confidence using various attacks from Cleverhans library, Mentor, Doctor Wei Yang, Fall 2017

1.     Used OCR model tesseract as our transfer learning model for this black-box attack

2.     Tried different attacks in perturbing the ID card to attack the model

 

 

Evasion Attack on SVM model that can detect malwares with high confidence, Mentor, Professor Carl Gunter and Doctor Wei Yang, Summer 2017-Fall 2017

1.     Built a comparative SVM model based on paper "Drebin: Effective and Explainable Detection of Android Malware in Your Pocket" (Copyright 2014 Internet Society, ISBN 1-891562-35-5)

2.     Evasion attack on the SVM model based on paper "Evasion Attacks Against Machine Learning at Test Time" (ECML PKDD, Part III, vol. 8190, LNCS, pp. 387--402. Springer, 2013)

 

Built a machine learning model that can automatically classify biomedical papers, City University of Hong Kong, Mentor, Professor Wong Ka Chun, Summer 2017

1.     Apply different machine learning models, including random forests, SVM, linear SVM and so on.

2.     Mainly focusing on improving current model's performance in imbalanced dataset.

Attack 2 layers' convolution network in recognizing hand-written digits, Promoting Undergraduate Research in Engineering(Pure) project, Mentor, Doctor Wei Yang, Spring 2017

1.     Take an example model from Tensorflow Tutorial (https://www.tensorflow.org/get_started/get_started )

2.     Attack the model using Saliency Map, based on the paper "The Limitations of Deep Learning in Adversarial Settings" (the 1st IEEE European Symposium on Security & Privacy, 2016)