Multivariate Time Series Classification with Crucial Timestamps Guidance

Our complete code will be made public after the paper is received.

💬 Introduction

This is CTNet model, the PyTorch source code of the paper "Multivariate Time Series Classification with Crucial Timestamps Guidance". Our method is a transformer-based method for Multivariate Time Series Classification. It has achieved the SOTA performance in the MTSC task on UEA archive.

📦UEA Dataset

Data is available at Google Drive .

You can also visit the official website .

📦Statistics of the Result

	TS2Vec	TNC	TS-TCC	TST	TapNet	MICOS	DKN	Formertime	FEAT	CTNet
ArticularyWordRecognition	0.987	0.973	0.953	0.977	0.987	0.99	0.993	0.984	0.991	0.987
AtrialFibrillation	0.2	0.133	0.267	0.067	0.333	0.333	0.467	0.6	0.293	1
BasicMotions	0.975	0.975	1	0.975	1	1	1	1	1	1
CharacterTrajectories	0.995	0.967	0.985	0.975	0.997	0.994	0.986	0.991	0.993	0.992
Cricket	0.972	0.958	0.917	1	0.958	1	0.951	0.981	0.969	1
DuckDuckGeese	0.68	0.46	0.38	0.62	0.575	0.74	0.56	0.6	0.564	0.76
EigenWorms	0.847	0.84	0.779	0.748	0.489	0.901	0.628	0.618	0.811	0.84
Epilepsy	0.964	0.957	0.957	0.949	0.971	0.971	0.979	0.964	0.948	0.964
ERing	0.874	0.852	0.904	0.874	0.133	0.941	0.933	0.904	0.896	0.944
EthanolConcentration	0.308	0.297	0.285	0.262	0.323	0.247	0.372	0.485	0.322	0.312
FaceDetection	0.501	0.536	0.544	0.534	0.556	0.523	0.631	0.687	0.53	0.646
FingerMovements	0.48	0.47	0.46	0.56	0.53	0.57	0.6	0.618	0.488	0.67
HandMovementDirection	0.338	0.324	0.243	0.243	0.378	0.649	0.662	0.567	0.378	0.567
Handwriting	0.515	0.249	0.498	0.225	0.357	0.621	0.231	0.214	0.542	0.399
Heartbeat	0.683	0.746	0.751	0.746	0.751	0.766	0.765	0.761	0.746	0.766
InsectWingbeat	0.466	0.469	0.264	0.105	0.208	0.218	0.362	0.227	0.462	0.322
JapaneseVowels	0.984	0.978	0.93	0.978	0.965	0.989	0.93	0.964	0.983	0.991
Libras	0.867	0.817	0.822	0.656	0.85	0.889	0.9	0.889	0.889	0.95
LSST	0.537	0.595	0.474	0.408	0.568	0.667	0.347	0.543	0.548	0.681
MotorImagery	0.51	0.5	0.61	0.5	0.59	0.5	0.62	0.632	0.562	0.59
NATOPS	0.928	0.911	0.822	0.85	0.939	0.967	0.872	0.961	0.921	0.939
PEMS-SF	0.682	0.699	0.734	0.74	0.751	0.809	0.948	0.774	0.874	0.855
PenDigits	0.989	0.979	0.974	0.56	0.98	0.981	0.93	0.981	0.989	0.99
PhonemeSpectra	0.233	0.207	0.252	0.085	0.175	0.276	0.525	0.147	0.216	0.133
RacketSports	0.855	0.776	0.816	0.809	0.868	0.941	0.879	0.842	0.888	0.947
SelfRegulationSCP1	0.812	0.799	0.823	0.754	0.652	0.799	0.913	0.887	0.852	0.852
SelfRegulationSCP2	0.578	0.55	0.533	0.55	0.55	0.578	0.6	0.592	0.562	0.578
SpokenArabicDigits	0.988	0.934	0.97	0.923	0.983	0.981	0.963	0.992	0.986	0.995
StandWalkJump	0.467	0.4	0.333	0.267	0.4	0.533	0.533	0.533	0.533	0.667
UWaveGestureLibrary	0.906	0.759	0.753	0.575	0.894	0.891	0.897	0.888	0.929	0.847
Total Best Acc.	0	1	1	1	2	6	8	3	2	14
Avg. Acc	0.704	0.670	0.667	0.617	0.657	0.742	0.732	0.727	0.722	0.772
Avg. Rank	5.42	7.47	7.32	8.25	5.78	3.87	4.53	4.5	4.73	3.13

🚀Network Architecture

👁️Requirements and Installation

• Python 11.2.0
• PyTorch 2.2.1
• NumPy 1.22.4
• cuda 12.2
• scikit-learn 1.0.2
• torchvision 0.17.1

🔍Start

To do.

📜Reference

To do.