Supporting pages for papers and projects in which I participated.
Ph.D dissertation.
In this work, we unfortunately discovered that several highly cited data mining papers have surprising problems with their proposed algorithm, datasets, or definition. To address these problems, we either provide potential solutions or offer recommendations.
Accpeted by IEEE ICDM2023 (Best Paper Runner-Up) & Springer KAIS.
In this work, we introduce C22MP, a novel anomaly detector, which combines catch22 and Matrix Profile to allow us to find subtle anomalies that no other SOTA algorithms can.
Accpeted by IEEE ICKG2022.
In this work, we introduce PRECIS, a novel distance measure for long time series, which exploits recent progress in our ability to summarize time series with “dictionaries”.
Accpeted by ACM SIGKDD2022 & Springer DMKD.
In this work, we introduce DAMP, a novel algorithm that allows us to find time series discords in datasets with trillions of datapoints for the first time, and to do it effectively online.
Accpeted by IEEE TKDE & IEEE ICDE2022.
In this work, we make a surprising claim: in spite of the fact that there are dozens of papers on early classification of time series, it is not clear that any of them could ever work in a real-world setting.
Accpeted by IEEE TKDE & IEEE ICDE2022.
In this work, we make a surprising claim: the majority of the individual exemplars in datasets created by Yahoo, Numenta, NASA, etc., suffer from one or more of four flaws.
Accpeted by IEEE TKDE & IEEE ICDE2021.
In this work, we make a surprising claim: in any realistic data mining application, the approximate FastDTW is orders of magnitude slower than the exact DTW.