[Retracted] Determination of Evidence Weights Based on Convolutional Neural Network for Classification Problem

Table 2

The classification performance comparison on the second type of datasets.

Dataset	Random state	MLP	FCN	ResNet	Dempster [17]	Deng [30]	Yan and Deng [33]	Han et al. [35]	Proposed method
ED	1	59.1%	69.2%	74.1%	71.4%	61.6%	61.6%	37.9%	76.5%
	2	59.2%	68.9%	74.2%	71.6%	61.8%	61.8%	37.3%	76.4%
	3	59.6%	69.6%	74.0%	71.9%	62.0%	62.0%	37.7%	77.1%
	4	59.8%	69.3%	74.2%	71.6%	62.3%	62.3%	37.2%	76.7%
	5	59.3%	69.4%	74.3%	71.4%	62.2%	62.2%	37.1%	76.4%
UY	1	68.1%	60.0%	64.1%	71.9%	62.8%	62.8%	70.4%	72.4%
	2	69.2%	59.9%	64.6%	72.0%	62.6%	62.6%	71.4%	73.5%
	3	68.0%	59.6%	64.4%	71.3%	62.5%	62.5%	71.1%	72.3%
	4	69.1%	60.3%	65.0%	72.5%	63.3%	63.3%	71.4%	73.7%
	5	69.2%	59.9%	64.4%	72.5%	63.1%	63.1%	71.2%	72.7%
UZ	1	68.3%	69.6%	75.8%	76.6%	70.1%	70.1%	64.5%	77.4%
	2	69.5%	69.4%	75.8%	76.9%	70.5%	70.5%	64.4%	77.5%
	3	68.8%	69.3%	75.8%	76.6%	70.2%	70.2%	64.0%	77.7%
	4	69.1%	69.9%	75.5%	76.4%	70.5%	70.5%	64.2%	76.5%
	5	68.4%	69.4%	75.2%	76.3%	70.4%	70.4%	64.2%	76.9%
CricketX	1	56.0%	79.1%	78.2%	72.1%	74.0%	74.0%	63.4%	80.4%
	2	54.8%	79.8%	78.5%	73.0%	72.7%	72.7%	63.4%	80.4%
	3	58.9%	80.7%	78.8%	73.7%	74.3%	74.3%	62.1%	80.8%
	4	57.0%	81.0%	79.4%	74.3%	73.7%	73.7%	62.1%	82.1%
	5	58.9%	80.7%	80.1%	73.0%	74.6%	74.6%	63.1%	81.4%
CricketY	1	58.6%	78.2%	78.5%	76.9%	68.9%	68.9%	70.1%	82.6%
	2	60.2%	79.4%	80.1%	77.8%	69.8%	69.8%	69.5%	83.4%
	3	60.8%	79.1%	79.4%	76.6%	70.1%	70.1%	69.8%	83.5%
	4	60.2%	78.5%	79.8%	76.6%	68.5%	68.5%	69.2%	81.6%
	5	57.6%	78.2%	79.1%	75.0%	68.5%	68.5%	70.8%	81.9%
CricketZ	1	61.2%	79.4%	80.1%	78.2%	73.3%	73.3%	69.5%	82.5%
	2	60.2%	80.4%	79.4%	77.2%	73.0%	73.0%	70.1%	81.4%
	3	60.2%	81.0%	80.4%	79.4%	74.3%	74.3%	69.8%	83.3%
	4	60.5%	79.1%	80.4%	78.2%	73.7%	73.7%	70.5%	82.6%
	5	61.8%	80.1%	81.7%	79.4%	75.3%	75.3%	70.8%	82.5%