arXiv cs.LG
· Papers
ITNet: A Learnable Integral Transform That Subsumes Convolution, Attention, and Recurrence
arXiv:2606.19538v3 Announce Type: replace-cross Abstract: Convolutional networks, recurrent networks, and transformers each encode different inductive biases -- locality, sequential memory, and content-dependent pairwise interaction -- and have remained mathematically distinct since their inception. We show that this f