BILCO: An Efficient Algorithm for Joint Alignment of Time Series
Authors: Xuelong Mi, Mengfan Wang, Alex Chen, Jing-Xuan Lim, Yizhi Wang, Misha B Ahrens, Guoqiang Yu
NeurIPS 2022 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We demonstrate the efficiency of BILCO using both synthetic and real experiments. Tested on thousands of datasets under various simulated scenarios and in three distinct application categories, BILCO consistently achieves at least 10 and averagely 20-folds increase in speed, and uses at most 1/8 and averagely 1/10 memory compared with the best existing max-flow method. |
| Researcher Affiliation | Academia | 1Dept. of Electrical and Computer Engineering, Virginia Tech 2Howard Hughes Medical Institute, Janelia Research Campus |
| Pseudocode | Yes | Algorithm 1 ELCO Algorithm 2 Relabel(R), R Gn |
| Open Source Code | Yes | Our source code can be found at https://github.com/yu-lab-vt/BILCO. |
| Open Datasets | Yes | Calculat[ing] signal propagation in imaging data (first two generated by us, last two from [20] and [21])... Extract depth information in binocular stereo (data from [16])... Signature identification (data from [14])... Part of the data are obtained from the coauthor, part of the data are public and cited. |
| Dataset Splits | No | The paper mentions using synthetic and real data for experiments but does not explicitly provide details about training, validation, and test dataset splits (e.g., percentages or sample counts). |
| Hardware Specification | Yes | All experiments were conducted in MATLAB with Intel(R) Xeon(R) Gold 6140@2.30Hz, 128GB memory, Windows 10 64-bit, and Microsoft VC++ compiler. No GPU is used. |
| Software Dependencies | No | The paper mentions 'MATLAB' and 'Microsoft VC++ compiler' but does not specify their version numbers or any other software dependencies with versions. |
| Experiment Setup | Yes | To make hyperparameter κ comparable, we normalized the synthetic data by dividing the standard deviation of the noise. We tested 20 instances for each combination of N, T, and κ. By setting initialization on each subgraph as the optimal warping path of the averaged subgraph... we set a window size with 1/5 sequence length in this experiment. |