For unseen ImageNet categories only, MLP-SGCZ slightly improves over the authors' GCNZ at both 2-hop and 3-hop accuracy with 4.3M versus 9.5M parameters, but remains below EXEM 1-nns and ADGPM, which use stronger visual features.

For unseen ImageNet categories only, MLP-SGCZ slightly improves over the authors' GCNZ at both 2-hop and 3-hop accuracy with 4.3M versus 9.5M parameters, but remains below EXEM 1-nns and ADGPM, which use stronger visual features. · CiteArk

测量结果

已评估 0/13 项

获得支持 0有争议或冲突 0尚未评估 13

保留同一结论下的完整结果组。已有评估排在前面；单项判断不替代整组结论。展开行查看条件、来源与历次证据。

每行是一项测量，分别显示论文报告、最近可用实测和累计证据状态。
指标与条件	论文报告	最近可用实测	证据判断
	27%	—	尚未评估
	7.1%	—	尚未评估
	26.6%	—	尚未评估
	6.3%	—	尚未评估
	19.8%	—	尚未评估
	4.1%	—	尚未评估
reportedPlusMinus: 0.2	20.9%	—	尚未评估
reportedPlusMinus: 0	4.3%	—	尚未评估
reportedPlusMinus: 0.2	21.2%	—	尚未评估
reportedPlusMinus: 0.1	4.4%	—	尚未评估
	9.5 millions of parameters	—	尚未评估
	4.3 millions of parameters	—	尚未评估

For unseen ImageNet categories only, MLP-SGCZ slightly improves over the authors' GCNZ at both 2-hop and 3-hop accuracy with 4.3M versus 9.5M parameters, but remains below EXEM 1-nns and ADGPM, which use stronger visual features.

报告指标与观测值

Assessment（0）

关联运行（0）

测量结果