notesum.ai

Published at November 21

Trajectory Representation Learning on Road Networks and Grids with Spatio-Temporal Dynamics

cs.AI

Released Date: November 21, 2024

Authors: Stefan Schestakov¹, Simon Gottschalk¹

Aff.: ¹L3S Research Center, Leibniz University Hannover, Hannover, Germany

Arxiv: http://arxiv.org/abs/2411.14014v1

		Task	Trajectory Similarity			Travel Time Estimation			Destination Prediction
		Metric	MR $\downarrow$	HR@1 $\uparrow$	HR@5 $\uparrow$	MAE $\downarrow$	MAPE $\downarrow$	RMSE $\downarrow$	F₁ $\uparrow$	Acc@1 $\uparrow$	Acc@5 $\uparrow$
Porto	Road	Transformer	7.604	0.170	0.562	152.83	23.87	222.43	0.084	0.174	0.390
		Trembr	13.74	0.284	0.572	162.32	25.24	236.73	0.118	0.228	0.498
		Toast	31.79	0.026	0.123	176.11	28.23	250.32	0.135	0.256	0.563
		JCLM	1.658	0.773	0.959	145.60	22.43	214.28	0.087	0.179	0.410
		LightPath	1.855	0.739	0.945	156.27	24.25	227.39	0.082	0.171	0.389
		START	2.636	0.808	0.944	129.05	19.73	192.08	0.075	0.158	0.375
	Grid	Transformer	8.134	0.210	0.576	107.78	16.10	168.59	0.081	0.171	0.407
		t2vec	33.10	0.103	0.273	122.65	18.55	185.20	0.084	0.174	0.416
		CLT-Sim	15.87	0.211	0.503	162.62	25.30	241.67	0.098	0.196	0.465
		CSTTE	5.378	0.710	0.820	110.91	16.89	169.59	0.085	0.175	0.423
		TrajCL	1.696	0.789	0.961	96.35	14.31	152.60	0.077	0.162	0.395
		TIGR	1.063	0.959	0.998	83.28	12.67	127.89	0.143	0.266	0.567
		Improvement	35.89%	18.69%	3.85%	13.57%	11.46%	16.19%	5.93%	3.91%	0.71%
San Francisco	Road	Transformer	6.027	0.336	0.727	119.01	22.06	184.98	0.039	0.079	0.218
		Trembr	17.82	0.193	0.450	121.88	23.00	188.16	0.040	0.081	0.218
		Toast	52.14	0.030	0.114	139.16	27.17	209.70	0.055	0.105	0.284
		JCLM	1.893	0.728	0.950	108.78	20.07	170.08	0.036	0.075	0.175
		LightPath	1.923	0.716	0.946	119.71	22.53	183.62	0.032	0.068	0.178
		START	4.153	0.747	0.867	111.83	21.04	172.00	0.030	0.061	0.175
	Grid	Transformer	10.69	0.218	0.543	80.12	14.66	133.60	0.038	0.078	0.228
		t2vec	29.19	0.122	0.308	108.32	20.55	164.48	0.051	0.101	0.287
		CLT-Sim	57.10	0.033	0.100	112.98	22.64	173.16	0.051	0.103	0.265
		CSTTE	7.612	0.394	0.704	74.33	14.36	118.25	0.041	0.085	0.231
		TrajCL	4.447	0.673	0.866	67.44	12.51	114.60	0.045	0.090	0.253
		TIGR	1.537	0.839	0.972	60.91	11.20	104.32	0.069	0.134	0.340
		Improvement	18.81%	12.32%	2.32%	9.68%	10.47%	8.97%	25.45%	27.62%	18.47%