Concept

STAC

A semi-supervised object detection method proposed in 2020 that uses a teacher network to generate pseudo-labels while a student network learns on strongly augmented images, representing early systematic exploration of consistency regularization for object detection.

Timeline (last 90 days)

Oct 12

In tests across four frontier agent models, X-AgentRed's harmful path discovery rate reached 1.77 times that of the baseline method STAC.

Unverified50%

All Facts (3)

Unverified

STAC（Securities Technology Analysis Center）成立于2006年，总部位于美国，是专门服务于金融服务业的独立技术基准测试机构

85%Unverified

STAC的成员涵盖高盛、摩根大通、芝加哥商品交易所等机构

80%Unverified

In tests across four frontier agent models, X-AgentRed's harmful path discovery rate reached 1.77 times that of the baseline method STAC.

50%