Efficient Learning of Timeseries Shapelets
Authors: Lu Hou, James Kwok, Jacek Zurada
AAAI 2016 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Extensive experimental results show that the proposed method is orders of magnitudes faster than the state-of-the-art shapelet-based methods, while achieving comparable or even better classification accuracy. |
| Researcher Affiliation | Academia | Lu Hou James T. Kwok Department of Computer Science and Engineering Hong Kong University of Science and Technology Hong Kong Jacek M. Zurada Department of Electrical and Computer Engineering University of Louisville, Louisville, KY, 40292, USA Information Technology Institute University of Social Science, Lodz 90-113, Poland |
| Pseudocode | Yes | Algorithm 1 FLAG. Input: timeseries of length Q, parameters α1, α2, ρ1, ρ2, stopping threshold ϵ. Output: shapelet indicator vector v. |
| Open Source Code | No | The codes (except UFS) are provided by authors of the various baseline methods, and the parameters are set as recommended. FSH is implemented in C++; IG, KW, FS, IGSVM and LTS are in Java; UFS and ours are in Matlab. |
| Open Datasets | Yes | Experiments are performed on the commonly used UCR1 and UEA2 data sets (Table 1) (Hills et al. 2014). 1http://www.cs.ucr.edu/ eamonn/time series data/. 2https://www.uea.ac.uk/computing/machine-learning/ shapelets/shapelet-data. |
| Dataset Splits | No | As is common in the shapelet literature (Grabocka et al. 2014; Grabocka, Wistuba, and Schmidt-Thieme 2015; Hills et al. 2014; Rakthanmanon and Keogh 2013), we use the default training/testing splits. |
| Hardware Specification | Yes | Experiments are performed on a desktop with Intel i7 CPU and 16GB memory. |
| Software Dependencies | No | The paper mentions software like C++, Java, and Matlab, but does not specify version numbers for any of these or any libraries/solvers. |
| Experiment Setup | No | The paper describes the ADMM solver and mentions parameters α1, α2, ρ1, ρ2, but it does not provide their specific values used in the experiments. It states that "the parameters are set as recommended" for baseline methods but not for their own proposed FLAG method. |