Table 22:

Complexity results for datasets in the HCI domain. U: unimodal models, M: multimodal fusion paradigms, O: optimization objectives, T: training structures.

	Dataset	ENRICO
	Metric	Epochs trained	Training time (s)	Training params (M)	Training peak memory (MB)	Inference time (s)	Inference params (M)
U	Unimodal $\left(i\right)$ Unimodal $\left(s\right)$	50 50	1601 1644	9.6 9.6	2796 2771	7.3 8.1	19.3 19.3
M	LF TF [179] LRTF [106] MI-Matrix [77]	50 50 50 50	1714 2012 1853 1604	19.3 19.3 19.3 19.3	2730 2718 2717 2730	8.7 10.9 9.7 8.5	19.3 19.3 19.3 19.3
O	CCA [145] RefNet [135]	50 50	2945 1747	19.3 25.7	2923 2757	9.1 13.8	19.3 25.7
T	GRadBlend [167]	50	2618	19.3	2610	12.1	19.3