Kong, Deguang
   Sr. Research Scientist  / Principal Engineer
Yahoo Research  / Yahoo

    701 First Avenue, Sunnyvale, 94089
    Phone:  408-718-4906
    Email:  doogkong  (AT)  gmail.com             

Dr. Deguang Kong is currently a Sr. Research Scientist at  Yahoo Research He previously  worked as a researcher at Samsung Research America, America Los Alamos National Lab and NEC Research Lab Silicon Valley. He has strong interdisciplinary background in machine learning/data mining and cyber security/privacy. He has worked on various research projects, including robust feature learning via structural sparsity, distance learning and label propagation for malware classification, and security-aware mobile app ranking and recommendation, etc. He has published over 20 referred articles in top conferences, including CCS; KDD, ICDM, SDM, WSDM, CIKM; ICML, NIPS, AAAI, ECML/PKDD, CVPR; SIGMETRICS, INFOCOM, etc. He was invited to serve as a reviewer for numerous top conferences and journals, such as KDD, ICDM, SDM, IJCAI, AAAI, TKDE, TNNLS, TIP, TCSVT, TKDD, DMKD, BMC Bioinformatics, and INFOCOM, IEEE transactions on dependable and secure computing (TDSC), IEEE TIFS,  IEEE Systems Journal, Winter Simulation Conference (WSC), CODASPY, SCN, etc.  

Dr. Kong's research spans both data mining/machine learning/big data analytics and cyber security and privacy. He held a Ph.D. in cyber security and privacy under the supervision of Dr. Peng Liu at the Cyber Security Lab Pennsylvania State University from Aug. 2008 to Aug. 2010, and worked on national cyber security projects (supported by DOE) at America Los Alamos National Lab from May 2012 to Dec. 2012. At Samsung Research America, he has been working on mobile cyber security/privacy and data mining/machine learning projects since Jan. 2014. His security paper entitled "SAS: semantic aware signature generation for polymorphic worm detection" was selected as one of three best papers in Securecomm'2010. His research on machine learning algorithms focuses on  robust feature learning and dimension reduction for big data, robust clustering and embedding in presence of noises, etc. His research work at NEC Research Lab Silicon Valley on exclusive feature learning on arbitrary structures was published at NIPS'2014. More information can be found at: Google scholarDBLP, LinkedIN, and Facebook.  His current works focus on big data,  security and privacy, etc.  

Work Experiences 

1. 2012, Los Alamos National Lab, Los Alamos,  NM (Intern)
2. 2013, NEC Research Lab, Cupertino, CA  (Intern)  
3. 2014--2016,  Samsung Research Lab, Mountain view, CA
-- Science driven innovations for mobile data science: algorithm, application and the lessons learned
4. 2016--Now,  Yahoo Research, Sunnyvale, CA

Research Interest

1. Malware,  Malicious app,  App User Engagement, Attack/defense models,  Adversarial learning,.....
2. Mutli-media Analytics:    A Privacy Perspective
3. Ads, Computational ads, Target Ads,  Fraud ads,  ... 
4. Shadow/Deep feature learning and applications

Research and Selected Publications 

1. Best paper of Securecomm'2010 (top 3 from 112)
2. Best paper of Samsung Incorporation'2015 (top 20 over 1200+ submission from all disciplines
  • A Holistic Approach towards Image Privacy (work at Samsung Research)
    • Pupples: Transformation supported personalized privacy preserving partial image sharing [ pdf J. He,  B. Liu D. Kong, etc,  etc,  DSN' 2016 
    • AUTOPRA: towards automatic personalized image privacy risk assessment for image sharing [ pdf ]                                                                                                                                                    D.Kong, P. Hu, J. Wang, etc   Samsung Technical Report' 2016 
  •  A Holistic Approach towards Mobile  Security and Privacy  (work at Samsung Research America)
    1. Malware Recomposition Attacks: exploiting feature confusions and evolutions in Android Apps. W Yang, D. Kong, etc,  Samsung Technical Report'2016 
    2. Personalized Mobile App Recommendation: Reconciling App Functionality and User Privacy Preference [ pdf ]
      Bin Liu, 
      Deguang Kong, Lei Cen, Neil Gong, Hongxia Jin and Hui Xiong, Proceedings of the Eighth International Conference on Web Search and Data Mining (WSDM'2015), Shanghai, China, February, 2015. (full paper, 16.4% acceptance). 
    3. Towards Permission Prediction on Mobile Apps via Structure Feature Learning [ pdf ]
      Deguang Kong
       and Hongxia Jin, Proceedings of 2015 SIAM International Conference on Data Mining (SDM'2015)  Canada, May, 2015. 
    4. Mobile App Security Risk Assessment:  A Crowdsourcing Ranking Approach from User Comments [ pdf 
      Lei Cen, 
      Deguang Kong, Hongxia Jin and Luo Si,  Proceedings of 2015 SIAM International Conference on Data Mining (SDM'2015)  Canada, May, 2015. 
    5. Protecting Your Children from Inappropriate Content in Mobile Apps: An Automatic Maturity Rating Framework [ pdf ]                                                                                                                          Bing Hu, Bin Liu, Neil Gong,  Deguang Kong and Hongxia Jin,  Proceedings of 24th ACM International Conference on Information and Knowledge Management (CIKM'2015)  Australia, Oct, 2015.  
    6. PinPlace: Associate Semantic Meanings with indoor Locations without Active Fingerprinting. [ pdf ] Xuan Bao, Bin Liu,  Bo Tang, Deguang Kong, Bing Hu and Hongxia Jin,  Proceedings of 24th ACM International Conference on Pervasive and Ubiquitous Computing (Ubicomp'2015),  Japan, Sep, 2015.   
    7. AUTOREB: Automatically Understanding Review-to-Behavior Fidelity in Android Apps [ pdf ]        Deguang Kong,  Lei Cen and Hongxia Jin,  Proceedings of 22nd ACM International Conference on Computer and Communications Security (CCS'2015),  Colorado, U.S,   Oct,  2015.  
  • Malware Classification and Attribution (work at Los Alamos National Lab)
    • Discriminant Malware Distance Learning on Structural Information for Automated Malware Classification. [ pdf ]
      Deguang Kong and Guanhua Yan,  
      Proceedings of the ACM International Conference on Measurement and Modeling of Computer Systems (SIGMETRICS'13 poster), Pittsburgh, PA, USA, June 2013.
    • Discriminant Malware Distance Learning on Structural Information for Automated Malware Classification. [ pdf ]
      Deguang Kong and Guanhua Yan. 
      Proceedings of the ACM International Conference on Knowledge Discovery and Data Mining (KDD'13), Chicago, IL, USA, August 2013.
    • Exploring Discriminatory Features for Automated Malware Classification. [ pdf ], 
      Guanhua Yan, Nathan Brown, and Deguang Kong. 
      Proceedings of the 10th Conference on Detection of Intrusions and Malware & Vulnerability Assessment (DIMVA'13), Berlin, Germany, July 2013.  
    • Transductive Malware Label Propagation: Find your lineage from your neighbors. [ pdf ]
      Deguang Kong and Guanhua Yan.  
      Proceedings of the IEEE Conference on Computer Communikations (INFOCOM'14), Toronto, Canada, April 2014.
  • Exploit Code Analysis via Semantic/Statistical Analysis (work at Penn State Univ. with Dr. Peng Liu)
    • SAS: Semantics Aware Signature Generation for Polymorphic Worm Detection [ pdf ]
      Deguang Kong
      , Yoon-chan Jhi, Tao Gong, Sencun Zhu, Peng Liu and Hongsheng Xi,  
      Proceedings of 6th International ICST Conference on Security and Privacy in Communication Networks (Securecomm'2010), Singapore, September, 2010  (best paper)
    • Extended Journal Version appeared at: International Journal of Information Security, 2011, Volume 10  
    • SA3: Automatic Semantic Aware Attribution Analysis of Remote Exploits [ pdf ]  
      Deguang Kong
      , Donghai Tian, Peng Liu and Dinghao Wu  
      Proceedings of 7th International ICST Conference on Security and Privacy in Communication Networks (Securecomm'2011), London, UK, September, 2011
    • Extended Journal version appeared at: Security and Communication Networks, 2013, 6(7)    
 Machine Learning   ( Subspace Learning -> Sparse Learning --> Deep Learning)
  • A Holistic Approach towards Deep Learning: Models, algorithms, applications, security and beyond 
    • Privacy-CNH: A framework to detect photo privacy with Convolutional Neural Network using Hierarchical Features     [ pdf ]                                                                                                                   L. Tran, D. Kong, J. Liu,  etc,  AAAI' 2016 
    •  Region-Sensitive Diversified Unsupervised Deep Feature Learning for Scientific Document Figure Classification   
  • Sparse Learning:  Exclusive Lasso  and Group Lasso (some works were done at NEC lab with Dr. Fujimaki)
    • Exclusive Feature Learning on Arbitrary Structures via L_{1,2}-norm [ pdf ]
      Deguang Kong, Ryohei Fujimaki, Ji Liu, etc,  
      Proceedings of Annual Conference on Neural Information processing System 2014 (NIPS'2014),  Montreal, Quebec, Canada, 2014
    • Uncorrelated Group LASSO  [ pdf ]
    • D. KongJi LiuBo LiuXuan Bao,  AAAI '20161765-1771
    • Exclusive Sparse Coding for multi-view Feature Learning 
    • D. Kong, etc.  
    • Efficient Algorithms for Selecting Features with Arbitrary Group Constraints via Group Lasso  [ pdf ]
    • D. Kong and C. Ding,  Proceedings of IEEE 13th  International Conference on Data Mining (ICDM'2013), Dallas, TX, Dec, 2013  
  • Subspace Learning:  feature selection, dimension reduction, semi-supervised learning, embedding
      • Pairwise Covariance Linear Discriminant Analysis Feature Learning  [ pdf ]
      • D. Kong and C. DingProceedings of 28th AAAI Conference on Artificial Intelligence, (AAAI'2014),  Quebec, Canada, July, 2014 
      • Multi-label Relief and F-statistic Feature Selection for Image Annotation [ pdf ]                             D. Kong, C. Ding, etc;Proceedings of IEEE 12th  International Conference on Computer Vision and Pattern Recognition  (CVPR'2012), Providence, RI, June, 2012  
      • A Semi-definite Positive Linear Discriminant Analysis and its Applications [ pdf ]    
        D. Kong and C. Ding,  
        Proceedings of IEEE 12th  International Conference on Data Mining (ICDM'2012), Brussels, Belgium, Dec, 2012  
      • An Iterative Locally Linear Embedding Algorithm  [ pdf ]  
        D.Kong and C. Ding, etc;
        Proceedings of  29th International Conference on Machine Learning  (ICML'2012),  Edinburgh, Scotland, TX, 2012 . 
      • Robust Non-negative Dictionary Learning [ pdf ] 
        Q. Pan, D. Kong, C. Ding, etc, 
        Proceedings of 28th AAAI Conference on Artificial Intelligence, (AAAI'2014),  Quebec, Canada, July, 2014 
      • Minimal Shrinkage for Noisy Data Recovery using Schatten-p Norm Objective [ pdf ]    
        D. Kong, M. Zhang and C. Ding;
        Proceedings of  European Conference on Machine Learning and Knowledge Discovery in Database (ECMLPKDD'2013), Prague, Czech Republic, Sep, 2013
      • Collective Kernel Construction in Noisy Environment   [ pdf ]  
        M. Zhang, C. Ding and D. Kong, 
        Proceedings of 2013 SIAM International Conference on Data Mining (SDM'2013),  Austin, Texas, May, 2013. 
      • Maximum Consistency Preferential Random Walks; [ pdf ]   
        D. Kong and C. Ding;
        Proceedings of  European Conference on Machine Learning and Knowledge Discovery in Database (ECMLPKDD'2012), Bristol, UK, Sep, 2012
      • Robust Non-negative Matrix Factorization using L_{2,1}-norm [ pdf ]   
        D. Kong, C. Ding and etc;
        Proceedings of  20th ACM Conference on Information and Knowledge Management (CIKM'2011), Glasgow, UK, Oct, 2011 

    I have keen interest in machine learning and security.  I am also open to any kind of collaborations. Contact me if interested!  

    Recent Services   
    Flag Counter