SARSA

Temporal difference learning for action values.

$$q_{t+1}(S_t,A_t)\leftarrow q_t(S_t,A_t)+\alpha \left(\stackrel{\text{TD error}}{\overbrace{\underset{\text{target}}{\underbrace{R_{t+1}+\gamma q_t(S_{t+1},A_{t+1})}}-q_t(S_t,A_t)}}\right)$$

It’s known as SARSA since it uses $(S_t,A_t,R_{t+1},S_{t+1},A_{t+1})$.