Share your thoughts, 1 month free Claude Pro on usSee more

Personality Recognition on AVI-6 (test)

0.1555MSE (H)

LLM-based AU-text fusion framework

Updated 25d ago

Evaluation Results

Method	Links
LLM-based AU-text fusion framework 2026.06		0.1555	0.138	0.1597	0.1077	0.1402	0.3242	0.2833	0.3287	0.2565	0.2982	0.4654	0.7258	0.5154	0.6934	0.6
Longformer (only text) 2026.06		0.1815	0.2697	0.2053	0.2097	0.2166	0.3508	0.4009	0.3721	0.3646	0.3721	0.3279	0.2736	0.0617	0.0022	0.1664
Longformer (AU semantic + text) 2026.06		0.2015	0.305	0.2058	0.2098	0.2305	0.3632	0.4183	0.3708	0.3613	0.3784	0.12	0.0196	0.0199	0.0533	0.0532
LSTM (Selected AUs) 2026.06		0.2057	0.299	0.2188	0.243	0.2416	0.3664	0.4286	0.3858	0.3985	0.3948	-0.0448	0.2236	-0.0343	0.1633	0.077
ViT (SW) 2026.06		0.2489	0.3551	0.2882	0.2785	0.2927	0.4084	0.4751	0.4145	0.4203	0.4296	0.1597	-0.041	-0.0131	0.1505	0.064
ResNet + BERT (feature fusion) 2026.06		0.256	0.2971	0.528	0.3146	0.3489	0.4036	0.4444	0.5769	0.4507	0.4689	0.1507	0.2573	0.1307	0.2537	0.1981
ResNet (DS) 2026.06		0.3743	0.5765	0.4058	0.5048	0.4654	0.4741	0.5987	0.5239	0.5823	0.5448	0.2014	-0.1573	0.1789	-0.0235	0.0499
ResNet (SW) 2026.06		0.4583	1.3539	0.9343	0.4982	0.8112	0.5436	0.947	0.8116	0.5777	0.7198	0.0387	-0.04	0.3675	-0.032	0.0836
ResNet + BERT (decision fusion) 2026.06		0.6673	0.5503	0.4162	0.4447	0.5196	0.6749	0.6139	0.5166	0.5272	0.5831	0.0215	0.1065	0.0957	0.2397	0.1159
Qwen-3-VL-8B-Instruct 2026.06		1.7777	0.4847	1.1193	1.1653	1.1368	1.2245	0.5349	0.9259	0.9574	0.9107	0.3814	0.6683	0.5589	0.5779	0.5467