Ryan Mercer

 

Ph.D. Candidate
Advisor: Eamonn Keogh
Data Mining and Machine Learning Lab
University of California, Riverside

I create time series tools. The Contrast Profile is one such tool which answers the question, “Which behaviors occur in A, but not in B?” This allows for identifying novel behaviors with next to zero labeling or parameter choices. Downstream applications include:

  • Data exploration

  • Preprocessing unstructured data for classification

  • Interpretable anomaly detection

I have industry experience in diverse domains:

  • Distracted driving detection at Toyota InfoTech Labs

  • Fraud detection at Visa Research

  • Neurological signal classification at UCLA Health, Neurology

  • Test design of hardware appliance at SanDisk

News

  • Novelets & MPlots papers accepted by ICDM (2022)

  • Distracted Driving demo published by SECON (2022)

  • Contrast Profile journal vers. published by DMKD (2022)

  • Contrast Profile paper published by ICDM (2021)

  • Payment network paper published by CIKM (2021)

  • 1NN DTW journal vers. published by DAMI (2021)

  • Toyota InfoTech internship (Jan. - Apr. 2021)

Publications

Matrix Profile XXV: Introducing Novelets: A Primitive that Allows Online Detection of Emerging Behavior in Time Series.
Ryan Mercer and Eamonn Keogh. Accepted by ICDM 2022.
[PDF] url

Matrix Profile XXVI: Mplots: Scaling Time Series Similarity Matrices to Massive Data.
Maryam Shahcheraghi, Ryan Mercer, João Manuel de Almeida Rodrigues, Audrey Der, Hugo Filipe Silveira Gamboa, Zachary Zimmerman and Eamonn Keogh. Accepted by ICDM 2022.
[PDF] url

Introducing the Contrast Profile: A Novel Time Series Primitive that Allows Real World Classification.
Ryan Mercer, Sara Alaee, Alireza Abdoli, Nader Shakibay Senobari, Shailendra Singh, Amy Murillo, and Eamonn Keogh. DMKD 2022.
[PDF] url

Matrix Profile XXIII: Contrast Profile: A Novel Time Series Primitive that Allows Real World Classification.
Ryan Mercer, Sara Alaee, Alireza Abdoli, Shailendra Singh, Amy Murillo, and Eamonn Keogh. ICDM 2021.
[PDF] url

Shape-Based Telemetry Approach for Distracted Driving Behavior Detection.
Ryan Mercer, Seyhan Ucar, and Eamonn Keogh. CSCN 2021.

Online Multi-horizon Transaction Metric Estimation with Multi-modal Learning in Payment Networks.
Chin-Chia Michael Yeh, Zhongfang Zhuang, Junpeng Wang, Yan Zheng, Javid Ebrahimi, Ryan Mercer, Liang Wang, and Wei Zhang. CIKM 2021.
[PDF]

Matrix Profile XXII: Exact Discovery of Time Series Motifs under DTW.
Sara Alaee, Ryan Mercer, Kaveh Kamgar, and Eamonn Keogh. DAMI 2021.
[PDF] url

Matrix Profile XXI: MERLIN: Parameter-Free Discovery of Arbitrary Length Anomalies in Massive Time Series Archives.
Takaaki Nakamura, Makoto Imamura, Ryan Mercer and Eamonn Keogh. ICDM 2020.
[PDF] url

Matrix Profile XX: Finding and Visualizing Time Series Motifs of All Lengths Using the Matrix Profile.
In 2019 IEEE International Conference on Big Knowledge (ICBK), pp. 175-182. IEEE, 2019.
[PDF] url

Intracranial Hypertension Detection using Machine Learning Classification of Non-invasive TCD Waveforms.
Fabien Scalzo, Ryan Mercer, Seth Wilk, Robert Hamilton. Military Health System Research Symposium, 2018.

Google Scholar Profile