AI RESEARCH
GradAttn: Replacing Fixed Residual Connections with Task-Modulated Attention Pathways
arXiv CS.CV
•
ArXi:2603.26756v1 Announce Type: new Deep ConvNets suffer from gradient signal degradation as network depth increases, limiting effective feature learning in complex architectures. ResNet addressed this through residual connections, but these fixed short-circuits cannot adapt to varying input complexity or selectively emphasize task relevant features across network hierarchies. This study