A DBN-Based Deep Neural Network Model with Multitask Learning for Online Air Quality Prediction

<figure class="figure-group responsive"><div class="sub-figure" id="a"><object data="https://static.hindawi.com/articles/jcse/volume-2019/5304535/figures/5304535.fig.006a.svgz" name="5304535.fig.006a" type="image/svg+xml"></object><div class="fig-label">(a) OL-MTL-DBN-DNN</div></div><div class="sub-figure" id="b"><object data="https://static.hindawi.com/articles/jcse/volume-2019/5304535/figures/5304535.fig.006b.svgz" name="5304535.fig.006b" type="image/svg+xml"></object><div class="fig-label">(b) OL-DBN-DNN</div></div><div class="sub-figure" id="c"><object data="https://static.hindawi.com/articles/jcse/volume-2019/5304535/figures/5304535.fig.006c.svgz" name="5304535.fig.006c" type="image/svg+xml"></object><div class="fig-label">(c) Winning-Model</div></div></figure>

<p>The prediction performances of different models for a 12-h horizon. In the pictures, time is measured along the horizontal axis and the concentrations of three kinds of air pollutants (<span class="nowrap"><svg height="11.6791pt" id="M49" style="vertical-align:-3.29112pt" version="1.1" viewbox="-0.0498162 -8.38798 30.632 11.6791" width="30.632pt" xmlns="http://www.w3.org/2000/svg"><g transform="matrix(.013,0,0,-0.013,0,0)"><path d="M46 650V622C120 617 128 613 128 525V125C128 42 120 34 40 28V0H311V28C221 34 212 39 212 124V281L286 262C297 261 316 261 331 263C429 275 526 338 526 468C526 533 501 579 462 609C422 638 364 650 293 650H46ZM212 559C212 588 215 600 223 606C230 613 251 618 279 618C361 618 430 572 430 464C430 337 350 302 285 302C252 302 225 309 212 314V559Z"></path></g><g transform="matrix(.013,0,0,-0.013,7.319,0)"><path d="M861 0V28C774 35 771 41 768 147L759 509C756 612 762 614 851 622V650H681L449 149L221 650H57V622C148 613 153 609 144 479L130 271C123 166 117 123 111 88C104 46 85 34 26 28V0H259V28C192 35 169 42 167 90C166 130 166 173 170 256L185 541H187L411 7H431L675 555H679L683 147C683 41 680 35 598 28V0H861Z"></path></g><g transform="matrix(.0091,0,0,-0.0091,18.973,3.132)"><path d="M414 144C384 79 371 75 317 75H135L276 221C367 316 408 376 408 465C408 570 327 635 237 635C179 635 131 609 100 575L42 494L67 471C94 510 138 565 205 565C277 565 321 517 321 435C321 348 258 270 195 195C146 137 88 81 33 26V0H411C423 44 433 88 446 135L414 144Z"></path></g><g transform="matrix(.0091,0,0,-0.0091,23.405,3.132)"><path d="M117 -12C153 -12 178 12 178 50C178 84 153 110 118 110C84 110 59 84 59 50C59 12 84 -12 117 -12Z"></path></g><g transform="matrix(.0091,0,0,-0.0091,25.561,3.132)"><path d="M158 548H390L417 615L410 623H122L83 318C105 326 143 337 185 337C296 337 350 275 350 188C350 116 308 42 225 42C164 42 122 74 100 93C90 101 82 99 72 92C60 82 51 68 50 59C48 46 52 38 66 24C82 9 125 -12 172 -12C225 -11 292 15 346 59C408 108 437 166 437 226C437 309 371 397 242 397C214 397 170 382 133 369L158 548Z"></path></g></svg>,</span> NO<sub>2</sub>, SO<sub>2</sub>) are measured along the vertical axis.</p>

Journal of Control Science and Engineering

fig6

Figure 6

Figure 6: A DBN-Based Deep Neural Network Model with Multitask Learning for Online Air Quality Prediction 