Model
OptNet
OptNet is a representative framework for embedding convex optimization problems as neural network layers. By integrating quadratic programming into the network architecture, it achieves end-to-end differentiable optimization with constraint satisfaction, and is considered a foundational work in the field of differentiable optimization layers.
Timeline (last 90 days)
Oct 12
Representative frameworks for differentiable optimization layers are OptNet and cvxpylayers, which treat convex optimization problems as a layer in a neural network, with gradients back-propagated via the implicit function theorem or KKT conditions.
Unverified50%