Youhao successfully defended his PhD thesis
In this thesis, we explored using learned compressive sensing for image compression using a deep neural network in the form of a deep equilibrium model (DEQ) as an alternative to the deep unrolling of more traditional iterative shrinkage-thresholding algorithm (ISTA). The advantage of the approach is a neural network that can speed up compressive sensing reconstruction using a fixed-point iterative approach and semi-tensor product (STP) decompositions cast in a neural network. Congratulations Youhao Yu! |