Yanping Chen

Ph.D Candidate in Computer Science and Engineering

Center for Research in Data Mining and Machine Learning

University of Callifornia, Riverside

Email: ychen053 [at] cs.ucr.edu
   

[Grant Awarded | Research | Publications | Professional Activities | Awards and Honors | Links]

About Me 
 
  • I am a fifth year Ph.D. student at the Dept. of Computer Science and Engineering at UCR.
  • I am working in the Data Mining Group. My advisor is TRUELY brilliant professor Dr. Eamonn Keogh.
  • My research interests lie in Data Mining, Machine Learning and Information Retrieval, with their applications to big data analysis, including times series data, vision data (videos/images), and audio data.
    • Grant Awarded  
     
  • I am the principal investigator on a $100,000 grant from Grand Challenges Explorations Round 11,funded by Bill & Melinda Gates Foundation.
    		•
         Our project, titled "Using Data to Understand Insect-Vectored Diseases", produces real-time information that can be used to plan effective suppression programs to combat problems such as malaria, by taking advantage of "The unreasonable effectiveness of data", obtained using our self-designed inexpensive sensors.
    Research
     
  •     Insect Detection and Classification Based on Wingbeat Sound     [project page].
    •  

    With an optical sensor, we can record on the order of millions of the “sound” of insect flights. The enormous amounts of data we collected allow us to take advantage of “The unreasonable effectiveness of data”. By exploring this huge dataset, we proposed a simple, but very accurate and robust classifier for insect classification.

     

    top) An audio snippet of a female Cx. stigmatosoma pursued by a male. bottom left) An audio snippet of a common house fly. bottom right) If we convert these sound snippets into periodograms, we can cluster and classify the insects.

     
  •     DTW-D: Time Series Semi-Supervised Learning from a Single Example     [pdf]   [slides] [project page].
    •  

    Direct application of off-the-shelf semi-supervised learning algorithms do not typically work well for time series. We explained why they typically fail, and we introduced a very simple but very effective fix. The fix requires only a single line of code, but it dramatically improves the performance of semi-supervised learning in time series problems.

     

    (a). Labeled dataset P and unlabeled dataset U; (b). Both ED/DTW failed to select the right object (U2) from U to add to P; (c) The proposed DTW-D succeeded to select U2;

     
  •     Towards Never-Ending Learning from Time Series Streams     [pdf] [project page].
    •  

    We proposed a framework for continuously discovering/learning patterns from time series data streams, with no/very little prior knowledge of the patterns to be learned. The data stream can be real-valued and never-ending. Our framework is very general and flexible. It is also scalable and robust to significant noise.

     

    (a). A pattern detected from a time series data stream, which was converted from an activity video. An oracle labeled it pushing and this pattern was added to the dictionary. (b) Another example of pushing was detected 9.6 minutes after it was discovered.

    Publications
     

    • Yanping Chen , Yuan Hao, Thanawin Rakthanmanon,Jesin Zakaria, Bing Hu, Eamonn Keogh. A General Framework for Never-Ending Learning from Time Series Streams. Journal of Data Mining and Knowledge Discovery 2014.     [pdf] [project page].

    • Francois Petitjean, Germain Forestier, Geoffrey I. Webb, Ann E. Nicholson, Yanping Chen , Eamonn Keogh. Dynamic Time Warping Averaging of Time Series allows Faster and more Accurate Classification. ICDM 2014 (Accepted, in Press).    

    • Yanping Chen , Adena Why, Gustavo Batista, Agenor Mafra-Neto, Eamonn Keogh. Flying Insect Classification with Inexpensive Sensors. Journal of Insect Behavior 2014.     [pdf] [project page].

    • Yanping Chen , Adena Why, Gustavo Batista, Agenor Mafra-Neto, Eamonn Keogh. Flying Insect Detection and Classification with Inexpensive Sensors. Journal of Visualized Experiments 2014 (Accepted, in Press).     [project page].

    • Yanping Chen ,Yuan Hao, Jesin Zakaria, Bing Hu, Thanawin Rakthanmanon, Eamonn Keogh. Towards Never-Ending Learning from Time Series Streams. SIGKDD 2013.     [pdf] [project page].

    • Yanping Chen, Bing Hu, Eamonn Keogh and Gustavo Batista. DTW-D: time series semi-supervised Learning from a Single Example. SIGKDD 2013.     [pdf] [project page].

    • Bing Hu, Yanping Chen, Jesin Zakaria, Liudmila Ulanova, and Eamonn Keogh. Classification of Multi-Dimensional Streaming Time Series by Weighting each Classifier’s Track Record. ICDM 2013.     [pdf]

    • Bing Hu, Yanping Chen and Eamonn Keogh. Time Series Classification under More Realistic Assumptions. SIAM SDM 2013.   [pdf] [slides] [project page]

    • Zhen Yu, Yanping Chen. A real-time motion detection algorithm for traffic monitoring systems based on consecutive temporal difference. ASCC 2009.   [pdf]

    Professional Activities
     
    • Invited demo in Solid Conference (as a Solid Fellow),  San Francisco, May 2014

    • Reviewer for:
          IEEE Transactions on Pattern Recognition and Machine Learning (TPAMI)
          IEEE Transactions on Knowledge and Data Engineering (TKDE)
          Springer Journal of Data Mining and Knowledge Discovery (DAMI)
    Awards and Honors
     
    • Grand Challenges Explorations Round 11 Grant Award
      by Bill & Melinda Gates Foundation

    • Solid Fellowship Award (2014)
      by Solid O'Reilly

    • Dissertation Year Program Fellowship Award (2014)
      by University of California, Riverside

    • SDM Student Travel Award (2014)
      by SDM 2014

    • Dean's Fellowship Award (2009-2011)
      by University of California, Riverside
    Links
     
    Last Updated: Oct 15th, 2014