(Time Series Domain Adaptation via Sparse Associative Structure Alignment)[AAAI2021]

Requirements

• Python 3.7

• Pytorch 1.7

Quick Start

You can run it with the following command.

python main.py -cuda_device 0 -dataset HHAR -batch_size 64 -seed 10 -epochs 40

result

HAR

dataset	domain	f1
HAR	12_to_16	84.91813
	18_to_27	99.61904667
	2_to_11	95.89705
	20_to_5	75.57394333
	24_to_8	90.73282333
	28_to_27	95.85423333
	30_to_20	88.80905333
	6_to_23	93.36903667
	7_to_13	92.67917
	9_to_18	80.68039667

HHAR

dataset	domain	f1
HHAR	0_to_2	73.22542667
	0_to_6	54.67753
	1_to_6	84.61155333
	3_to_8	77.30236333
	4_to_5	86.79520333
	5_to_0	42.24310333
	6_to_1	88.1413
	7_to_4	89.41448667
	8_to_3	83.05941

WISDM

dataset	domain	f1
WISDM	17_to_23	51.51952333
	2_to_11	82.16593333
	20_to_30	84.00848
	23_to_32	51.45548
	28_to_4	90.03818
	33_to_12	81.55814
	35_to_31	74.76046333
	5_to_26	48.03187
	6_to_19	70.86559333
	7_to_18	72.58147333