Xutong Ren (任旭彤)
Hi, I am Xutong Ren, a graduate student majoring in Machine Learning at Carnegie Mellon University, Pittsburgh, US. My research interests lie primarily in the area of Computer Vision and Machine Learning, specifically, style transfer, image recognition and weakly supervised learning. I am looking for a summer intership in 2020. Welcome to contact me if you are interested. [CV]
- Carnegie Mellon University (CMU)
- Peking University (PKU)
2019 - 2020, Department of Machine Learning, School of Computer Science (SCS)
Master of Science in Machine Learning
2015 - 2019, The School of Electronics Engineering and Computer Science (EECS)
Bachelor of Science in Computer Science and Technology
- Research Assistant
- Research Intern
- Machine Learning Competition
- Teaching Assistant
- Winter Visitor
Google AI ML Winter Camp, Google, Beijing
Introduction to Computer Systems, EECS, Peking University
Robotics Program, Massachusetts Institute of Technology
Chen Wei, Lingxi Xie, Xutong Ren, Yingda Xia, Chi Su, Jiaying Liu, Qi Tian, Alan Yuille, "Iterative Reorganization with Weak Spatial Constraints: Solving Arbitrary Jigsaw Puzzles for Unsupervised Representation Learning," in IEEE Conference on Computer Vision and Pattern Recognition (CVPR), June 2019.[Paper]   [Code]
Xutong Ren, Mading Li, Wen-Huang Cheng and Jiaying Liu, "Joint Enhancement and Denoising Method via Sequential Decomposition," in IEEE International Symposium on Circuits and Systems (ISCAS), May 2018, pp. 1-5. (oral)[Paper]   [Project]   [Code]   [Slide]
Xutong Ren, Lingxi Xie, Chen Wei, Siyuan Qiao, Chi Su, Jiaying Liu, Qi Tian, Elliot Fishman and Alan Yuille, "Generalized Coarse-to-Fine Visual Recognition with Progressive Training".
Xutong Ren, Wenhan Yang, Wen-Huang Cheng, Jiaying Liu, “LR3M: Robust Low-Light Enhancement via Low-Rank Regularized Retinex Model,” submitted to IEEE Transactions on Image Processing (TIP).
Enhancement and Denoising Method via Sequential Decomposition
In this work we solved the problem that low-light enhancement methods ignore intensive noise in original images which leads to simultaneously enhance the noise as well. A joint low-light enhancement and denoising strategy based on a novel sequential Retinex decomposition concept was proposed. Our method performs well for a wide variety of images, and achieves better quality compared with the state-of-the-art methods. This work was done during my internship at ICST and is accepted by ISCAS 2018.
Joint Low-Light and Low-Rank Method
In this work we explained and demonstrated why solving the Retinex decomposition problem iteratively causes noise to distribute in each component, which in the end impairs the denoising ability. We are the first to combine low-light enhancement method and low-rank denoising approach in order to improve the quality of images. An integrated low-rank decomposition of three channels in RGB space is applied to remove noise from the reflectance map.
This work was done during my internship at ICST and is submitted to TIP.
Coarse-to-Fine with Progressive Training
In this work we proposed a generalized coarse-to-fine framework that can bring gains in a wide range of visual recognition tasks, like classification, localization and segmentation. A training strategy named progressive training is designed, which follows an organized learning order by gradually reducing additional information given to the fine stage and increasing the learning difficulty. This work was done during my summer visit to JHU and it is under review now.
Transfer Learning via Jigsaw Puzzle
In this work we built a recurrent solution to a jigsaw puzzle of arbitrary different permutations in a self-supervised manner, without heavy burden of human annotation. By enforcing the neural network to learn from spatial contexts of puzzles, we were capable to transfer the learned features to visual tasks such as medical image analyses. It can solve more difficult puzzles and provide help in training 3D data. This work was done during my summer visit to JHU and it is accepted by CVPR 2019.
Google AI ML Winter Camp
In this project, we focused on the domain of image to image translation. Although there are already many supervised methods perform really well, but they rely heavily on annotations. On the other hand, some unsupervised methods can successfully transfer an image to other domains, but they all focus on whole image transfer, not capable of specific attributes transfer. Therefore, we wanted to realize face attribute transfer on real human images in an unsupervised way, using cartoon images as a bridge.
Teaching Assistant of Itroduction to Computor Systems (ICS)
Introduction to Computer Systems is a core course of computer science. Both Peking University and Carnegie Mellon University start this course (click to see the course information). In 2017-2018 I was selected to become one of the teaching asistants. My job was to teach in class, review projects and answer questions. One of my student had got the highest score in the exam and also it is the highest in the course's history.
I participated in an education program held by Peking University and Massachusetts Institute of Technology in Jan. 2018 - Feb.2018, which was a robotic programming. I manually assembled a robot and then programmed on it. I coded a program to guide the robot to go along the road in a printed map from a randomly set start location to a corresponding final location, using all kinds of sensors.
I am a member of Young Volunteers Association of EECS, Peking University. I have done a lot of volunteer jobs, including computer repair service once a month, teaching in hope schools and helping in international holdings. I was thus awarded as excellent volunteer of Peking University. Besides, when I was in America, I volunteered in the Food Bank, distributing food to people.
I was the associate minister of Ministry of Social Practice of EECS, Peking University. I organized some social practices such as summer practice and business visit. The group I leaded went to Hangzhou, China to investigate two kinds of intangible cultural heritages: the silk and paper cutting. In 2015 my group was ranked Best Summer Practice Group in Peking University.
- Most Technical Project Award for Google AI ML Winter Camp (2019)
- Peking University Award for Academic Excellents (2018)
- Wang Shengdi Scholarship (2018, top 10%)
- Peking University Award for Academic Excellents (2017)
- 8108 College Scholarship (2017, top 10%)
- The Third Prize of Peking University ACM ICPC (2017)
- Peking University Award for Excellent Volunteers (2016)