Table 1 Hyperparameters.
From: Efficient crowd simulation in complex environment using deep reinforcement learning
Parameter | Value | Description |
|---|---|---|
\(\gamma\) | 0.99 | Discount factor |
\(A_rl\) | 1e–4 | Actor learning rate |
\(C_rl\) | 1e–4 | Critic learning rate |
\(B_{size}\) | 256 | Batch size |
\(RB_{size}\) | 1e6 | Reply buffer size |
\(\tau\) | 0.005 | Target update rate |
\(P_n\) | 0.2 | Policy noise |
\(E_n\) | 0.25 | Exploration noise |
\(w_1\) | 0.15 | Â |
\(w_2\) | 0.08 | Â |
\(w_3\) | 1.2 | Â |
\(w_4\) | 0.4 | Â |
\(r_{arrv}\) | 10 | Â |
\(r_{move}\) | – 0.6 |  |
