A Dynamic Adjusting Reward Function Method for Deep Reinforcement Learning with Adjustable Parameters

<div>Accumulated reward per episode of DQN (different <span class="nowrap"><svg height="9.39034pt" id="M116" style="vertical-align:-3.42943pt" version="1.1" viewbox="-0.0498162 -5.96091 6.54569 9.39034" width="6.54569pt" xmlns="http://www.w3.org/2000/svg" xmlns:xlink="http://www.w3.org/1999/xlink"><g transform="matrix(.013,0,0,-0.013,0,0)"><path d="M471 401C471 442 455 448 440 448C421 448 389 439 366 424C314 390 233 331 165 242H163L187 345C193 372 197 393 197 409C197 435 189 448 174 448C146 448 80 408 23 349L40 327C64 351 97 373 107 373C115 373 118 366 111 337L29 -4L36 -12C56 -4 83 3 110 9C123 69 136 126 150 172C227 283 336 381 377 381C392 381 396 370 379 298L326 74C290 -77 273 -172 273 -197C273 -219 316 -261 346 -261L353 -246C348 -227 352 -174 374 -66L457 310C467 352 471 381 471 401Z"></path></g></svg>).</span></div>

Mathematical Problems in Engineering

fig9

Figure 9

Figure 9: A Dynamic Adjusting Reward Function Method for Deep Reinforcement Learning with Adjustable Parameters 