Computer Science Information Systems

Chengcheng Jia August **, ****

**** ******** ** ******: 1-857-***-****

Boston, MA, USA, 02215

E-mail: ac1urs@r.postjobfree.com Homepage:chengchengjia Education

Northeastern University Boston, MA

Ph.D. Major: Computer Engineering 2013 { 2016

{ Department of Electronic and Computer Engineering (Full Scholarship, 3 years)

Jilin University Changchun, China

Ph.D. Major: Computer Science 2010 { 2013

{ Department of Computer Science and Technology (Full Scholarship, 3 years)

Jilin University Changchun, China

M.Sc. Major: Computer Science 2007 { 2010

{ Department of Computer Science and Technology (Full Scholarship, 3 years)

Northeast Normal University Changchun, China

B.Sc. Major: Software Engineering 2003 { 2007

{ Department of Computer Science and Technology

Work Experience

TVision Insights Boston, MA

Computer Vision Engineering 04/2017 { present

{ Face detection and expression recognition: We installed Kinect cameras in 86 households, to capture RGB and infrered human faces then recognize expressions, in order to recommend productions to them. I lead the vision part, including face detection and expression recognition, using OpenCV, KinectSDK toolboxes, AWS.

Paci c Northwest National Laboratory Richland, WA Intern, Supervisor: Abhinav Vishnu 08/2016 { 10/2016

{ Visual representation prediction for unlabeled video: We aim to predict the visual representation of unlabeled video, by LSTM framework. The previous works usually make use of the current frames to predict the future, however, we consider that the deep further frames with rich knowledge are also useful for prediction. Therefore, we design a bi-direction mechanism for prediction. In the rst stage, I used CNN on Ca e model, adding an extra k-clustering layer as the last-second layer for unlabeled prediction. In the next stage, I would design an LSTM model compared with CNN.

FX Palo Alto Laboratory, CA, U.S. Palo Alto, CA

Intern, Supervisor: Patrick Chiu, Qiong Liu, Yanying Chen, Sven Kratz 05/2015 { 08/2015

{ Gesture-object interaction: We aim to predict the gesture-object interactions during a tele-conference. We collect a new dataset with di erent gestures and objects, and design a deep learning framework for prediction of a long video. Research Interests

Machine learning:

{ Action recognition application: RGBD human action, gesture-object interaction

{ Transfer learning, low-rank learning, supervised/unsupervised learning

Numerical Analysis:

{ I work on Tensor decomposition for 6+ years, including Tucker decomposition, Tensor-Train decomposition, Non-Negative Tensor factorization, Low-Rank tensor completion.

Deep learning:

{ Denoising auto-encoder, CNN, Long-Short-Term-Memory for visual representation Research Experience

Northeastern University Boston, MA

Research Assistant, Supervisor: Prof. Yun (Raymond) Fu 2013 { 2016

{ Canonical Temporal Alignment: We aim to solve three challenges in action recognition: multi-subject, multi-view, and sub-action. We rst select key frames from action sequences, then align the key frames temporally to eliminate the variance of the same class, nally project the samples to a common tensor subspace for action recognition.

{ Low-Rank Tensor for Action Recognition: We make use of low-rank tensor completion to nd a common subspace for action recognition. Furthermore, we suppose all the samples are distributed in a Remannian manifold, and design a low-rank decomposition mechanism from the manifold.

{ Deep Tensor Decomposition: Designed a deep decomposition model with non-negative tensor factorization (NTF) and tensor-train (TT) decomposition. Proposed a subspace learning method by tensor completion on a Riemannian manifold.

{ Tensor Transfer Learning: Proposed a latent transfer model to solve missing modality problem, we deliver missing depth modality from an RGBD source domain to an RGB target domain for action recognition.

{ Deep Auto-Encoder: Designed a deep tensor auto-encoder model for spatiotemporal corrupted action recognition, to address large corruptions and uncertain ratios problems. Designed LSTM based algorithm for action recognition and prediction.

Jilin University Changchun, China

Research Assistant, Supervisor: Prof. Zhezhou Yu 2010 { 2013

{ The research on group action recognition in complicated context (Grant No.20121103). Built the human group database towards group action recognition. Proposed an e cient model in complicated environment to parse interaction of group in a video.

{ Research of Group Action Recognition Technology (Grant No. 201115022). Built the human group videos database towards group activities understanding, including pedestrian tracking, people counting, and group action classi cation. Publications (available at publications)

Peer-Reviewed Journal Articles

12. Chengcheng Jia, Ming Shao, Yun Fu. "Sparse Canonical Temporal Alignment with Deep Tensor Decomposition for Action Recognition". Transactions on Image Processing, 2016. 11. Chengcheng Jia, Yun Fu. "Low-Rank Tensor Subspace Learning for RGB-D Action Recognition". Transactions on Image Processing, 2016. 10. Yu, Zhe-Zhou, Yu-Hao Liu, Bin Li, Shu-Chao Pang, and Cheng-Cheng Jia. "Incremental Graph Regulated Nonnegative Matrix Factorization for Face Recognition." Journal of Applied Mathematics, 2014

9. Cheng-Cheng Jia, Su-Jing Wang, Xu-Jun Peng, Wei Pang, Can-Yan Zhang, Chun-Guang Zhou, Zhe-Zhou Yu, "Incremental Multi-linear Discriminant Analysis Using Canonical Correlations for Action Recognition", Neurocomputing, 2012.

8. Zhe-Zhou Yu, Cheng-Cheng Jia, Wei Pang, Can-Yan Zhang, Li-Hua Zhong, "Incremental Multi-linear Discriminant Analysis Using Canonical Correlations for Action Recognition", IEEE Signal Processing Letters, 2012.

7. Cheng-Cheng Jia, Xiang-Li Xu, Chun-Guang Zhou, Cai-Tang Sun, Li-Biao Zhang, "Human Skeleton Modeling Using Chain Codes", Journal of Jilin University: Natural Science Edition, 2010. 6. Yuhao Liu, Chengcheng Jia, Bin Li,Shuchao Pang, Zhezhou Yu, "Graph Regularized Projective Non-negative Matrix Factorization for face recognition", Journal of Computational Information Systems, 2013.

5. Yuhao Liu, Chengcheng Jia, Bin Li, Zhezhou Yu, "Gradient descent Fisher Non-negative Matrix Factorization for face recognition" Journal of Information & Computational Science, 2013. 4. Yuhao Liu, Chengcheng Jia, Bin Li, Erping Pang, Zhezhou Yu, "A New Method to Construct Neighbor Graph for Non-negative Matrix Factorization on Manifold", Journal of Information & Computational Science, 2013.

3. Yuhao Liu, Chengcheng Jia, Bin Li, Shuchao Pang, Zhezhou Yu, "A Modi ed Subclass Discriminant Non-negative Matrix Factorization with Projected Gradient Descent",Journal of Computational Information Systems, 2013.

2. Rui Liu, Cheng-Cheng Jia, Zhe-Zhou Yu, "Global and Local Information Based Spherical Marginal Fisher Analysis for Face Recognition", Journal of Information & Computational Science, 2013.

1. Qing Wen, Cheng-Cheng Jia, Yang-Quan YU, Gang Chen, Zhe-Zhou YU, Chun-Guang Zhou,

"People Number Estimation in the Crowded Scenes Using Texture Analysis Based on Gabor Filter", Journal of Computational Information Systems, 2011. Conference Proceedings

6. Chengcheng Jia, Ming Shao and Yun Fu. Sparse Canonical Temporal Alignment with NTF for RGB-D Action Recognition, IJCNN 2016.

5. Shuhui Jiang, Ming Shao, Chengcheng Jia, Yun Fu. Consensus Style Centralizing Auto-encoder for Weak Style Classi cation, AAAI 2016.

4. Chengcheng Jia, Guoqiang Zhong, YunFu. Low-Rank Tensor Learning with Discriminant Analysis for Action Classi cation and Image Recovery,AAAI, 2014. (Oral paper) 3. Chengcheng Jia, Yu Kong, Zhengming Ding, YunFu. Latent Tensor Transfer Learning for RGB-D Action Recognition,ACM MM, 2014. (Long paper) 2. Chengcheng Jia, Wei Pang, Yun Fu. Mode-driven volume analysis based on correlation of time series, VECTaR 2014.

1. Cheng-Cheng Jia, Su-Jing Wang, Chun-Guang Zhou, Cai-Tang Sun, Li-Biao Zhang "Tensor analysis and multi-scale features based multi-view human action recognition", Computer Engineering and Technology (ICCET), 2010.

Book Chapters

3. Chengcheng Jia and Yun Fu. Subspace Learning for Action Recognition, Human Activity Recognition and Prediction, Springer, pages 49{69, 2016. 2. Chengcheng Jia, Wei Pang and Yun Fu. Multimodal Action Recognition, Human Activity Recognition and Prediction, Springer, pages 71{85, 2016. 1. Chengcheng Jia, Yu Kong, Zhengming Ding and Yun Fu. RGB-D Action Recognition, Human Activity Recognition and Prediction, Springer, pages 87{106, 2016. Papers Under Review

2. Chengcheng Jia, Zhengming Ding, Yu Kong, Yun Fu. Cross-Modality Action Recognition by Latent Tensor Transfer Learning. IEEE Transactions on Circuits and Systems for Video Technology (TCSVT). Major revision.

1. Chengcheng Jia, Ming Shao, Sheng Li, Handong Zhao, Yun Fu. Stacked Denoising Tensor Auto-Encoder for Action Recognition with Spatiotemporal Corruptions. TIP. Major revision. Presentations

Latent Tensor Transfer Learning for RGB-D Action Recognition The 22nd ACM International Conference on Multimedia (ACM MM) November 2014 Awards, Grants & Honours

ACM MM Travel grant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2014 National Scholarship of China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2013 Third Prize of the 27th Postgraduate Elite Cup of Jilin University . . . . . . . . . . . . . . . 2013 First Prize of the 26th Postgraduate Elite Cup of Jilin University . . . . . . . . . . . . . . . 2012 Technical Skills

Operation System: Windows, Linux

Ca e, TensorFlow

Programming Languages

{ Pro cient: C++, Python, Matlab

{ Familiar with: OpenCV, Java, HTML

Contact this candidate