TSN

KO | EN

gitlite — search

#javascript #python #react #llm #go #android #machine-learning #rust #deep-learning #java #awesome #kubernetes #awesome-list #openai

(★ 14)

The paper "Triple-shapelet Networks for Time Series Classification"

TSN 최신버젼 다운로드

최종 버전 다운로드 (.zip)

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# TSN
The paper "Triple-shapelet Networks for Time Series Classification" accepted by 2019 IEEE International Conference on Data Mining (ICDM).

## Code
The code for Triple-shapelet Networks model.

### Dependencies
- Keras 2.0 and above
- tensorflow 1.0 and above

### Usage
This repository contains a demo of ECGFiveDays in the UCR dataset.  The ECGFiveDays dataset is located in `UCR_TS_Archive_2015/ECGFiveDays/`. You can run the command
```
python triple_shapelet_network_demo.py
```
to test the model.

## Classification Result
Here is the full result of TSN on 85 UCR time series classification datasets. The results of the deep learning methods are taken from [1], and the results of the other methods are from [2].

| Dataset | DDDTW | DTDC | BOSS | TSF | TSBF | LPS | EE | COTE | MLP | FCN | ResNet | Encoder | TSN |
|:--|--|--|--|--|--|--|--|--|--|--|--|--|--|
|Adiac | 0.701 | 0.701 | 0.765 | 0.731 | 0.770 | 0.770 | 0.665 | 0.790 | 0.397 | **0.844** | 0.829 | 0.484 | 0.798 |
|ArrowHead | 0.789 | 0.720 | 0.834 | 0.726 | 0.754 | 0.783 | 0.811 | 0.811 | 0.778 | 0.843 | 0.845 | 0.804 | **0.869** |
|Beef | 0.667 | 0.667 | 0.800 | 0.767 | 0.567 | 0.600 | 0.633 | 0.867 | 0.720 | 0.697 | 0.753 | 0.643 | **0.900** |
|BeetleFly | 0.650 | 0.650 | **0.900** | 0.750 | 0.800 | 0.800 | 0.750 | 0.800 | 0.870 | 0.860 | 0.850 | 0.745 | **0.900** |
|BirdChicken | 0.850 | 0.800 | 0.950 | 0.800 | 0.900 | **1.000** | 0.800 | 0.900 | 0.775 | 0.955 | 0.885 | 0.665 | 0.800 |
|Car | 0.800 | 0.783 | 0.833 | 0.767 | 0.783 | 0.850 | 0.833 | 0.900 | 0.767 | 0.905 | **0.925** | 0.758 | 0.917 |
|CBF | 0.997 | 0.980 | 0.998 | 0.994 | 0.988 | **0.999** | 0.998 | 0.996 | 0.872 | 0.994 | 0.995 | 0.947 | 0.989 |
|Chlorine | 0.708 | 0.713 | 0.661 | 0.720 | 0.692 | 0.608 | 0.656 | 0.727 | 0.802 | 0.814 | 0.844 | 0.573 | **0.868** |
|CinCECGtorso | 0.725 | 0.852 | 0.887 | 0.983 | 0.712 | 0.736 | 0.942 | **0.995** | 0.840 | 0.824 | 0.826 | 0.911 | 0.975 |
|Coffee | **1.000** | **1.000** | **1.000** | 0.964 | **1.000** | **1.000** | **1.000** | **1.000** | 0.996 | **1.000** | **1.000** | 0.979 | **1.000** |
|Computers | 0.716 | 0.716 | 0.756 | 0.720 | 0.756 | 0.680 | 0.708 | 0.740 | 0.563 | **0.822** | 0.815 | 0.574 | 0.624 |
|CricketX | 0.754 | 0.754 | 0.736 | 0.664 | 0.705 | 0.697 | **0.813** | 0.808 | 0.591 | 0.792 | 0.791 | 0.694 | 0.721 |
|CricketY | 0.777 | 0.774 | 0.754 | 0.672 | 0.736 | 0.767 | 0.805 | **0.826** | 0.600 | 0.787 | 0.803 | 0.675 | 0.723 |
|CricketZ | 0.774 | 0.774 | 0.746 | 0.672 | 0.715 | 0.754 | 0.782 | **0.815** | 0.617 | 0.811 | 0.812 | 0.692 | 0.723 |
|Diatom | 0.967 | 0.915 | 0.931 | 0.931 | 0.899 | 0.905 | 0.944 | 0.928 | 0.910 | 0.313 | 0.301 | 0.913 | **0.990** |
|DistPhxAgeGp | 0.705 | 0.662 | 0.748 | 0.748 | 0.712 | 0.669 | 0.691 | 0.748 | 0.657 | 0.710 | 0.717 | 0.737 | **0.835** |
|DistPhxCorr | 0.732 | 0.725 | 0.728 | 0.772 | 0.783 | 0.721 | 0.728 | 0.761 | 0.726 | 0.760 | 0.771 | 0.741 | **0.810** |
|DistPhxTW | 0.612 | 0.576 | 0.676 | 0.669 | 0.676 | 0.568 | 0.647 | 0.698 | 0.617 | 0.690 | 0.665 | 0.688 | **0.785** |
|Earthquakes | 0.705 | 0.705 | 0.748 | 0.748 | 0.748 | 0.640 | 0.741 | 0.748 | 0.717 | 0.727 | 0.712 | 0.748 | **0.820** |
|ECG200 | 0.830 | 0.840 | 0.870 | 0.870 | 0.840 | 0.860 | 0.880 | 0.880 | 0.916 | 0.889 | 0.874 | **0.923** | 0.900 |
|ECG5000 | 0.924 | 0.924 | 0.941 | 0.939 | 0.940 | 0.917 | 0.939 | 0.946 | 0.929 | 0.940 | 0.934 | 0.940 | **0.948** |
|ECGFiveDays | 0.769 | 0.822 | **1.000** | 0.956 | 0.877 | 0.879 | 0.820 | 0.999 | 0.970 | 0.987 | 0.975 | 0.982 | **1.000** |
|ElectricDevices | 0.592 | 0.594 | **0.799** | 0.693 | 0.703 | 0.681 | 0.663 | 0.713 | 0.592 | 0.702 | 0.729 | 0.674 | 0.625 |
|FaceAll | 0.902 | 0.899 | 0.782 | 0.751 | 0.744 | 0.767 | 0.849 | 0.918 | 0.793 | **0.945** | 0.839 | 0.793 | 0.811 |
|FaceFour | 0.830 | 0.818 | **1.000** | 0.932 | **1.000** | 0.943 | 0.909 | 0.898 | 0.840 | 0.928 | 0.955 | 0.815 | 0.955 |
|FacesUCR | 0.904 | 0.908 | **0.957** | 0.883 | 0.867 | 0.926 | 0.945 | 0.942 | 0.833 | 0.946 | 0.955 | 0.874 | 0.940 |
|FiftyWords | 0.754 | 0.754 | 0.705 | 0.741 | 0.758 | 0.818 | **0.820** | 0.798 | 0.684 | 0.627 | 0.740 | 0.723 | 0.690 |
|Fish | 0.943 | 0.926 | **0.989** | 0.794 | 0.834 | 0.943 | 0.966 | 0.983 | 0.848 | 0.958 | 0.979 | 0.866 | 0.943 |
|FordA | 0.723 | 0.765 | 0.930 | 0.815 | 0.850 | 0.873 | 0.738 | **0.957** | 0.730 | 0.904 | 0.920 | 0.923 | 0.933 |
|FordB | 0.667 | 0.653 | 0.711 | 0.688 | 0.599 | 0.711 | 0.662 | 0.804 | 0.603 | 0.878 | 0.913 | 0.890 | **0.918** |
|GunPoint | 0.980 | 0.987 | **1.000** | 0.973 | 0.987 | 0.993 | 0.993 | **1.000** | 0.927 | **1.000** | 0.991 | 0.936 | 0.967 |
|Ham | 0.476 | 0.552 | 0.667 | 0.743 | 0.762 | 0.562 | 0.571 | 0.648 | 0.691 | 0.718 | 0.757 | 0.727 | **0.781** |
|HandOutlines | 0.868 | 0.865 | 0.903 | 0.919 | 0.854 | 0.881 | 0.889 | **0.919** | 0.918 | 0.806 | 0.911 | 0.899 | 0.831 |
|Haptics | 0.399 | 0.399 | 0.461 | 0.445 | 0.490 | 0.432 | 0.393 | **0.523** | 0.433 | 0.480 | 0.519 | 0.427 | 0.438 |
|Herring/Otoliths | 0.547 | 0.547 | 0.547 | 0.609 | 0.641 | 0.578 | 0.578 | 0.625 | 0.528 | 0.608 | 0.619 | 0.586 | **0.688** |
|InlineSkate | **0.562** | 0.509 | 0.516 | 0.376 | 0.385 | 0.500 | 0.460 | 0.495 | 0.337 | 0.339 | 0.373 | 0.292 | 0.340 |
|InsWngSnd | 0.355 | 0.473 | 0.523 | 0.633 | 0.625 | 0.551 | 0.595 | **0.653** | 0.607 | 0.393 | 0.507 | 0.633 | 0.624 |
|ItalyPower | 0.950 | 0.951 | 0.909 | 0.960 | 0.883 | 0.923 | 0.962 | 0.961 | 0.954 | 0.961 | 0.963 | 0.965 | **0.969** |
|LrgKitApp | 0.795 | 0.795 | 0.765 | 0.571 | 0.528 | 0.717 | 0.811 | 0.845 | 0.473 | **0.902** | 0.900 | 0.619 | 0.707 |
|Lightning2 | 0.869 | 0.869 | 0.836 | 0.803 | 0.738 | 0.820 | **0.885** | 0.869 | 0.670 | 0.739 | 0.770 | 0.692 | 0.787 |
|Lightning7 | 0.671 | 0.658 | 0.685 | 0.753 | 0.726 | 0.740 | 0.767 | 0.808 | 0.630 | 0.827 | **0.845** | 0.625 | 0.767 |
|Mallat | 0.949 | 0.927 | 0.938 | 0.919 | 0.960 | 0.908 | 0.940 | 0.954 | 0.918 | 0.967 | **0.972** | 0.876 | 0.940 |
|Meat | 0.933 | 0.933 | 0.900 | 0.933 | 0.933 | 0.883 | 0.933 | 0.917 | 0.897 | 0.853 | **0.968** | 0.742 | 0.933 |
|MedicalImages | 0.737 | 0.745 | 0.718 | 0.755 | 0.705 | 0.746 | 0.742 | 0.758 | 0.721 | **0.779** | 0.770 | 0.734 | 0.722 |
|MidPhxAgeGp | 0.539 | 0.500 | 0.545 | 0.578 | 0.578 | 0.487 | 0.558 | 0.636 | 0.531 | 0.553 | 0.569 | 0.579 | **0.798** |
|MidPhxCorr | 0.732 | 0.742 | 0.780 | **0.828** | 0.814 | 0.773 | 0.784 | 0.804 | 0.770 | 0.801 | 0.809 | 0.761 | 0.778 |
|MidPhxTW | 0.487 | 0.500 | 0.545 | 0.565 | 0.597 | 0.526 | 0.513 | 0.571 | 0.534 | 0.512 | 0.484 | 0.592 | **0.637** |
|MoteStrain | 0.833 | 0.768 | 0.879 | 0.869 | 0.903 | 0.922 | 0.883 | **0.937** | 0.858 | **0.937** | 0.928 | 0.840 | 0.908 |
|NonInvThor1 | 0.806 | 0.841 | 0.838 | 0.876 | 0.842 | 0.812 | 0.846 | 0.931 | 0.916 | **0.956** | 0.945 | 0.916 | 0.877 |
|NonInvThor2 | 0.893 | 0.890 | 0.901 | 0.910 | 0.862 | 0.841 | 0.913 | 0.946 | 0.917 | **0.953** | 0.946 | 0.932 | 0.902 |
|OliveOil | 0.833 | 0.867 | 0.867 | 0.867 | 0.833 | 0.867 | 0.867 | **0.900** | 0.667 | 0.723 | 0.830 | 0.400 | **0.900** |
|OSULeaf | 0.880 | 0.884 | 0.955 | 0.583 | 0.760 | 0.740 | 0.806 | 0.967 | 0.557 | 0.977 | **0.979** | 0.576 | 0.711 |
|PhalCorr | 0.739 | 0.761 | 0.772 | 0.803 | 0.830 | 0.756 | 0.773 | 0.770 | 0.735 | 0.820 | **0.839** | 0.767 | 0.823 |
|Phoneme | 0.269 | 0.268 | 0.265 | 0.212 | 0.276 | 0.237 | 0.305 | **0.349** | 0.096 | 0.325 | 0.334 | 0.172 | 0.216 |
|Plane | **1.000** | **1.000** | **1.000** | **1.000** | **1.000** | **1.000** | **1.000** | **1.000** | 0.978 | **1.000** | **1.000** | 0.976 | **1.000** |
|ProxPhxAgeGp | 0.800 | 0.795 | 0.834 | 0.849 | 0.849 | 0.795 | 0.805 | 0.854 | 0.856 | 0.831 | 0.853 | 0.844 | **0.859** |
|ProxPhxCorr | 0.794 | 0.794 | 0.849 | 0.828 | 0.873 | 0.842 | 0.808 | 0.869 | 0.733 | 0.903 | **0.921** | 0.791 | 0.880 |
|ProxPhxTW | 0.766 | 0.771 | 0.800 | **0.815** | 0.810 | 0.732 | 0.766 | 0.780 | 0.767 | 0.767 | 0.780 | 0.812 | **0.815** |
|RefDev | 0.445 | 0.445 | 0.499 | **0.589** | 0.472 | 0.459 | 0.437 | 0.547 | 0.379 | 0.508 | 0.525 | 0.488 | 0.445 |
|ScreenType | 0.429 | 0.437 | 0.464 | 0.456 | 0.509 | 0.416 | 0.445 | 0.547 | 0.403 | **0.625** | 0.622 | 0.383 | 0.459 |
|ShapeletSim | 0.611 | 0.600 | **1.000** | 0.478 | 0.961 | 0.867 | 0.817 | 0.961 | 0.503 | 0.724 | 0.779 | 0.530 | 0.872 |
|ShapesAll | 0.850 | 0.838 | 0.908 | 0.792 | 0.185 | 0.873 | 0.867 | 0.892 | 0.771 | 0.895 | **0.921** | 0.758 | 0.812 |
|SmlKitApp | 0.640 | 0.648 | 0.725 | **0.811** | 0.672 | 0.712 | 0.696 | 0.776 | 0.371 | 0.783 | 0.786 | 0.596 | 0.709 |
|SonyAIBOSurf1 | 0.742 | 0.710 | 0.632 | 0.787 | 0.795 | 0.774 | 0.704 | 0.845 | 0.672 | **0.960** | 0.958 | 0.743 | 0.857 |
|SonyAIBOSurf2 | 0.892 | 0.892 | 0.859 | 0.810 | 0.778 | 0.872 | 0.878 | 0.952 | 0.834 | **0.979** | 0.978 | 0.839 | 0.909 |
|StarlightCurves | 0.962 | 0.962 | 0.978 | 0.969 | 0.977 | 0.963 | 0.926 | **0.980** | 0.949 | 0.961 | 0.972 | 0.957 | 0.947 |
|Strawberry | 0.954 | 0.957 | 0.976 | 0.965 | 0.954 | 0.962 | 0.946 | 0.951 | 0.961 | 0.972 | **0.981** | 0.946 | 0.977 |
|SwedishLeaf | 0.901 | 0.896 | 0.922 | 0.914 | 0.915 | 0.920 | 0.915 | 0.955 | 0.851 | **0.969** | 0.956 | 0.930 | 0.915 |
|Symbols | 0.953 | 0.963 | **0.967** | 0.915 | 0.946 | 0.963 | 0.960 | 0.964 | 0.832 | 0.955 | 0.906 | 0.821 | 0.949 |
|SyntheticControl | 0.993 | 0.997 | 0.967 | 0.987 | 0.993 | 0.980 | 0.990 | **1.000** | 0.976 | 0.985 | 0.998 | 0.996 | 0.997 |
|ToeSegmentation1 | 0.807 | 0.807 | 0.939 | 0.741 | 0.781 | 0.877 | 0.829 | **0.974** | 0.583 | 0.961 | 0.963 | 0.659 | 0.921 |
|ToeSegmentation2 | 0.746 | 0.715 | **0.962** | 0.815 | 0.800 | 0.869 | 0.892 | 0.915 | 0.745 | 0.880 | 0.906 | 0.795 | 0.877 |
|Trace | **1.000** | 0.990 | **1.000** | 0.990 | 0.980 | 0.980 | 0.990 | **1.000** | 0.807 | **1.000** | **1.000** | 0.960 | **1.000** |
|TwoLeadECG | 0.978 | 0.985 | 0.981 | 0.759 | 0.866 | 0.948 | 0.971 | 0.993 | 0.762 | **1.000** | **1.000** | 0.863 | 0.972 |
|TwoPatterns | **1.000** | **1.000** | 0.993 | 0.991 | 0.976 | 0.982 | **1.000** | **1.000** | 0.946 | 0.871 | **1.000** | 1.000 | 0.962 |
|UWaveGestX | 0.779 | 0.775 | 0.762 | 0.804 | **0.831** | 0.829 | 0.805 | 0.822 | 0.767 | 0.754 | 0.780 | 0.786 | 0.798 |
|UWaveGestY | 0.716 | 0.698 | 0.685 | 0.727 | 0.736 | **0.761** | 0.726 | 0.759 | 0.698 | 0.639 | 0.670 | 0.696 | 0.706 |
|UWaveGestZ | 0.696 | 0.679 | 0.695 | 0.743 | **0.772** | 0.768 | 0.724 | 0.750 | 0.697 | 0.726 | 0.750 | 0.711 | 0.698 |
|UWaveGestAll | 0.935 | 0.938 | 0.939 | 0.957 | 0.926 | 0.966 | **0.968** | 0.964 | 0.955 | 0.817 | 0.860 | 0.954 | 0.952 |
|Wafer | 0.980 | 0.993 | 0.995 | 0.996 | 0.995 | 0.997 | 0.997 | **1.000** | 0.996 | 0.997 | 0.999 | 0.996 | 0.999 |
|Wine | 0.574 | 0.611 | 0.741 | 0.630 | 0.611 | 0.630 | 0.574 | 0.648 | 0.565 | 0.587 | 0.744 | 0.500 | **0.870** |
|WordSynonyms | 0.730 | 0.730 | 0.638 | 0.647 | 0.688 | 0.755 | **0.779** | 0.757 | 0.598 | 0.564 | 0.622 | 0.613 | 0.658 |
|Worms | 0.584 | 0.649 | 0.558 | 0.610 | 0.688 | 0.701 | 0.662 | 0.623 | 0.457 | 0.765 | **0.791** | 0.571 | 0.591 |
|WormsTwoClass | 0.649 | 0.623 | **0.831** | 0.623 | 0.753 | 0.753 | 0.688 | 0.805 | 0.601 | 0.726 | 0.747 | 0.639 | 0.696 |
|Yoga | 0.856 | 0.856 | **0.918** | 0.859 | 0.819 | 0.869 | 0.879 | 0.877 | 0.855 | 0.839 | 0.870 | 0.820 | 0.843 |
| best | 5 | 3 | 15 | 5 | 5 | 5 | 8 | 19 | 0 | 17 | 15 | 1 | **24** |
| AVG rank | 8.676 | 8.800 | 6.082 | 7.259 | 7.329 | 7.635 | 6.835 | **3.529** | 10.312 | 5.641 | 4.465 | 8.994 | 5.441 |

## Reference
[1] Fawaz H I, Forestier G, Weber J, et al. Deep learning for time series classification: a review[J]. Data Mining and Knowledge Discovery, 2019, 33(4): 917-963. 
[2] Bagnall A, Lines J, Bostrom A, et al. The great time series classification bake off: a review and experimental evaluation of recent algorithmic advances[J]. Data Mining and Knowledge Discovery, 2017, 31(3): 606-660.

TSN

TSN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

TSN

TSN 최신버젼 다운로드

오늘의 추천 프로젝트

Was this content helpful?

Recent Feedback

Download README