Residual Algorithms
Mentioned in Deep Residual Reinforcement Learning to explain the difference.
It’s essentially the idea of doing backprop with the target network too.
Residual: “I’ll try to fully minimize the Bellman error, including how my parameters affect the next-state value.”