注冊(cè) | 登錄讀書好,好讀書,讀好書!
讀書網(wǎng)-DuShu.com
當(dāng)前位置: 首頁(yè)出版圖書科學(xué)技術(shù)計(jì)算機(jī)/網(wǎng)絡(luò)人工智能基于深度學(xué)習(xí)的道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)

基于深度學(xué)習(xí)的道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)

基于深度學(xué)習(xí)的道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)

定 價(jià):¥98.00

作 者: 崔建勛 等
出版社: 電子工業(yè)出版社
叢編項(xiàng):
標(biāo) 簽: 暫缺

ISBN: 9787121430190 出版時(shí)間: 2022-04-01 包裝:
開本: 16開 頁(yè)數(shù): 296 字?jǐn)?shù):  

內(nèi)容簡(jiǎn)介

  這本書系統(tǒng)闡述了深度學(xué)習(xí)方法論在道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)領(lǐng)域的**研究成果。需要著重說(shuō)明以下幾點(diǎn):(1)領(lǐng)域限定在了道路交通,因?yàn)榻煌ㄊ莻€(gè)大系統(tǒng),存在著航空、水運(yùn)、道路等多種運(yùn)輸方式,而本書所闡述的研究均是針對(duì)道路交通領(lǐng)域的數(shù)據(jù)以及面向道路交通領(lǐng)域的應(yīng)用;(2)本書所討論的研究問(wèn)題是道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)問(wèn)題,該問(wèn)題是時(shí)空數(shù)據(jù)挖掘領(lǐng)域中時(shí)空預(yù)測(cè)問(wèn)題的一個(gè)重要子集,在本書的第1章中將會(huì)對(duì)這個(gè)問(wèn)題進(jìn)行數(shù)學(xué)上的形式化定義;(3)本書針對(duì)道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)問(wèn)題的討論,完全是基于深度學(xué)習(xí)的方法論,所參考的文獻(xiàn)絕大部分發(fā)表于2017年以后,并不涵蓋前人對(duì)該研究問(wèn)題所采用的全部方法論(如ARIMA,卡爾曼濾波、SVR等)。

作者簡(jiǎn)介

  崔建勛,哈爾濱工業(yè)大學(xué) 交通學(xué)院 副教授碩士生導(dǎo)師,長(zhǎng)期從事人工智能與道路交通的交叉領(lǐng)域研究,主要研究方向包括基于深度學(xué)習(xí)的短時(shí)交通狀態(tài)預(yù)測(cè)、基于深度強(qiáng)化學(xué)習(xí)的自動(dòng)駕駛決策規(guī)劃與控制等。

圖書目錄

目\t錄
第 1 章\t道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)總論.................................................... 001
1.1 時(shí)空數(shù)據(jù)............................................................................................................... 001
1.2 時(shí)空數(shù)據(jù)挖掘....................................................................................................... 002
1.3 道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè) ....................................................................... 003
1.3.1 問(wèn)題描述 .................................................................................................. 003
1.3.2 核心挑戰(zhàn) .................................................................................................. 005
1.3.3 問(wèn)題分類 .................................................................................................. 007
1.4 道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)研究概要性綜述 ........................................... 012
1.5 基于深度學(xué)習(xí)的道路短期交通狀態(tài)時(shí)空序列預(yù)測(cè)建模一般性框架................ 014
1.6 本章小結(jié)............................................................................................................... 015
第 1 篇\t基于深度學(xué)習(xí)的網(wǎng)格化道路交通狀態(tài)時(shí)空序列預(yù)測(cè)
第 2 章\t基于 2D 圖像卷積神經(jīng)網(wǎng)絡(luò)的時(shí)空相關(guān)性建模................................... 018
2.1 ST-ResNet ............................................................................................................. 020
2.1.1 問(wèn)題提出 .................................................................................................. 020
2.1.2 歷史交通狀態(tài)切片數(shù)據(jù)的獲取............................................................... 020
2.1.3 預(yù)測(cè)模型 .................................................................................................. 022
2.1.4 訓(xùn)練算法 .................................................................................................. 026
2.2 MDL...................................................................................................................... 027
2.2.1 問(wèn)題提出 .................................................................................................. 027
2.2.2 預(yù)測(cè)模型 .................................................................................................. 029
2.2.3 訓(xùn)練算法 .................................................................................................. 035
2.3 MF-STN ................................................................................................................ 036
2.3.1 問(wèn)題提出 .................................................................................................. 037
2.3.2 預(yù)測(cè)模型 .................................................................................................. 037
2.3.3 訓(xùn)練算法 .................................................................................................. 040
2.4 DeepLGR[23] .......................................................................................................... 042
2.4.1 問(wèn)題提出 .................................................................................................. 043
2.4.2 預(yù)測(cè)模型 .................................................................................................. 043
2.4.3 模型小結(jié) .................................................................................................. 048
2.5 ST-NASNet ........................................................................................................... 048
2.5.1 問(wèn)題提出 .................................................................................................. 051
2.5.2 預(yù)測(cè)模型 .................................................................................................. 051
2.5.3 訓(xùn)練算法 .................................................................................................. 054
2.6 本章小結(jié)............................................................................................................... 055
第 3 章\t基于 2D 圖像卷積與循環(huán)神經(jīng)網(wǎng)絡(luò)相結(jié)合的時(shí)空相關(guān)性建模....... 057
3.1 STDN[25]................................................................................................................ 058
3.1.1 問(wèn)題提出 .................................................................................................. 059
3.1.2 預(yù)測(cè)模型 .................................................................................................. 059
3.1.3 訓(xùn)練算法 .................................................................................................. 066
3.2 ACFM[26] ............................................................................................................... 067
3.2.1 問(wèn)題提出 .................................................................................................. 067
3.2.2 預(yù)測(cè)模型 .................................................................................................. 068
3.2.3 模型拓展 .................................................................................................. 073
3.2.4 訓(xùn)練算法 .................................................................................................. 075
3.3 PredRNN[27] .......................................................................................................... 076
3.4 PredRNN++[28] ...................................................................................................... 081
3.4.1 模型架構(gòu) .................................................................................................. 082
3.4.2 Casual-LSTM............................................................................................ 083
3.4.3 GHU.......................................................................................................... 084
3.5 MIM[29].................................................................................................................. 084
3.6 SA-ConvLSTM[30]................................................................................................. 088
3.6.1 模型背景 .................................................................................................. 089
3.6.2 模型構(gòu)造 .................................................................................................. 090
3.7 本章小結(jié)............................................................................................................... 092
第 4 章\t基于 3D 圖像卷積的時(shí)空相關(guān)性建模..................................................... 094
4.1 問(wèn)題提出............................................................................................................... 095
4.2 預(yù)測(cè)模型............................................................................................................... 095
4.2.1 近期時(shí)空相關(guān)性捕獲模塊....................................................................... 096
4.2.2 短期時(shí)空相關(guān)性捕獲模塊....................................................................... 098
4.2.3 特征融合模塊........................................................................................... 099
4.2.4 預(yù)測(cè)模塊 .................................................................................................. 099
4.2.5 損失函數(shù) .................................................................................................. 099
4.3 訓(xùn)練算法............................................................................................................... 100
4.4 本章小結(jié)............................................................................................................... 100
第 2 篇\t基于深度學(xué)習(xí)的拓?fù)浠缆方煌顟B(tài)時(shí)空序列預(yù)測(cè)
第 5 章\t基于 1D 圖像卷積與卷積圖神經(jīng)網(wǎng)絡(luò)相結(jié)合的時(shí)空相關(guān)性建模 .. 102
5.1 STGCN[32] ............................................................................................................. 102
5.1.1 問(wèn)題提出 .................................................................................................. 102
5.1.2 模型建立 .................................................................................................. 103
5.2 TSSRGCN[33] ........................................................................................................ 105
5.2.1 問(wèn)題提出 .................................................................................................. 106
5.2.2 模型建立 .................................................................................................. 106
5.3 Graph Wave Net[34]................................................................................................ 112
5.3.1 問(wèn)題提出 .................................................................................................. 112
5.3.2 模型建立 .................................................................................................. 113
5.4 ASTGCN[35] .......................................................................................................... 116
5.4.1 問(wèn)題提出 .................................................................................................. 116
5.4.2 模型建立 .................................................................................................. 117
5.5 本章小結(jié)............................................................................................................... 123
第 6 章\t基于循環(huán)與卷積圖神經(jīng)網(wǎng)絡(luò)相結(jié)合的時(shí)空相關(guān)性建模.................... 124
6.1 AGC-Seq2Seq[36]................................................................................................... 124
6.1.1 問(wèn)題提出 .................................................................................................. 125
6.1.2 模型建立 .................................................................................................. 125
6.2 DCGRU[37] ............................................................................................................ 129
6.2.1 問(wèn)題提出 .................................................................................................. 130
6.2.2 模型建立 .................................................................................................. 130
6.3 T-MGCN[38] ........................................................................................................... 132
6.3.1 問(wèn)題提出 .................................................................................................. 132
6.3.2 模型建立 .................................................................................................. 133
6.4 GGRU[39] ............................................................................................................... 138
6.4.1 符號(hào)定義 .................................................................................................. 139
6.4.2 GaAN 聚合器 ........................................................................................... 140
6.4.3 GGRU 循環(huán)單元 ...................................................................................... 141
6.4.4 基于 Encoder-Decoder 架構(gòu)和 GGRU 的交通狀態(tài)時(shí)空預(yù)測(cè)網(wǎng)絡(luò) ........ 141
6.5 ST-MetaNet[40]....................................................................................................... 142
6.5.1 問(wèn)題提出 .................................................................................................. 143
6.5.2 模型建立 .................................................................................................. 143
6.6 本章小結(jié)............................................................................................................... 147
第 7 章\t基于 Self-Attention 與卷積圖神經(jīng)網(wǎng)絡(luò)相結(jié)合的時(shí)空相關(guān)性建模.... 149
7.1 GMAN[41] .............................................................................................................. 150
7.1.1 問(wèn)題提出 .................................................................................................. 150
7.1.2 模型建立 .................................................................................................. 150
7.2 ST-GRAT[42] .......................................................................................................... 157
7.2.1 問(wèn)題提出 .................................................................................................. 157
7.2.2 模型建立 .................................................................................................. 158
7.3 STTN[43] ................................................................................................................ 163
7.3.1 問(wèn)題提出 .................................................................................................. 163
7.3.2 模型建立 .................................................................................................. 164
7.4 STGNN[44] ............................................................................................................. 169
7.4.1 問(wèn)題提出 .................................................................................................. 169
7.4.2 模型建立 .................................................................................................. 169
7.5 本章小結(jié)............................................................................................................... 173
第 8 章\t基于卷積圖神經(jīng)網(wǎng)絡(luò)的時(shí)空相關(guān)性同步建模 ...................................... 174
8.1 MVGCN[45] ........................................................................................................... 175
8.1.1 問(wèn)題提出 .................................................................................................. 176
8.1.2 模型建立 .................................................................................................. 177
8.2 STSGCN[46] ........................................................................................................... 180
8.2.1 問(wèn)題提出 .................................................................................................. 180
8.2.2 模型建立 .................................................................................................. 180
8.3 本章小結(jié)............................................................................................................... 186
第 3 篇\t深度學(xué)習(xí)相關(guān)基本理論
第 9 章\t全連接神經(jīng)網(wǎng)絡(luò) ............................................................................................. 190
9.1 理論介紹............................................................................................................... 190
9.2 本章小結(jié)............................................................................................................... 192
第 10 章\t卷積神經(jīng)網(wǎng)絡(luò) ............................................................................................... 193
10.1 二維卷積神經(jīng)網(wǎng)絡(luò)(2D CNN)....................................................................... 193
10.2 一維卷積和三維卷積神經(jīng)網(wǎng)絡(luò)(1D 和 3D CNN) ........................................ 198
10.3 擠壓和激勵(lì)卷積網(wǎng)絡(luò)(Squeeze and Excitation Networks)............................ 199
10.4 殘差連接網(wǎng)絡(luò)(ResNet) ................................................................................. 201
10.5 因果卷積(Casual CNN)................................................................................. 202
10.6 膨脹卷積(Dilated Convolution) .................................................................... 203
10.7 可變形卷積(Deformable Convolution) ......................................................... 204
10.8 可分離卷積(Separable Convolution) ............................................................ 206
10.9 亞像素卷積(SubPixel Convolution).............................................................. 207
10.10 本章小結(jié)........................................................................................................... 208
第 11 章\t循環(huán)神經(jīng)網(wǎng)絡(luò)................................................................................................ 210
11.1 標(biāo)準(zhǔn)循環(huán)神經(jīng)網(wǎng)絡(luò)(RNN)............................................................................. 211
11.2 雙向循環(huán)神經(jīng)網(wǎng)絡(luò)(Bi-RNN)........................................................................ 211
11.3 深度循環(huán)神經(jīng)網(wǎng)絡(luò)(Deep RNN) ................................................................... 212
11.4 長(zhǎng)短期記憶神經(jīng)網(wǎng)絡(luò)(LSTM)[60] .................................................................. 213
11.5 門控循環(huán)單元(GRU)..................................................................................... 215
11.6 ConvLSTM ......................................................................................................... 216
11.7 本章小結(jié)............................................................................................................. 217
第 12 章\t卷積圖神經(jīng)網(wǎng)絡(luò)........................................................................................... 218
12.1 譜域圖卷積[66] .................................................................................................... 220
12.1.1 拓?fù)鋱D數(shù)據(jù)上的卷積操作推導(dǎo)............................................................. 220
12.1.2 切比雪夫多項(xiàng)式卷積............................................................................. 225
12.1.3 圖卷積網(wǎng)絡(luò)(Graph Convolutional Networks,GCN)....................... 226
12.1.4 擴(kuò)散卷積(Diffusion Convolution)..................................................... 226
12.2 空間域圖卷積..................................................................................................... 228
12.2.1 頂點(diǎn)域圖卷積特征聚合器的一般性定義 ............................................. 228
12.2.2 GraphSAGE[71]........................................................................................ 229
12.2.3 GAT......................................................................................................... 232
12.3 本章小結(jié)............................................................................................................. 235
第 13 章\t注意力機(jī)制(Attention)......................................................................... 236
13.1 Encoder-Decoder 模型[75-77] ................................................................................ 236
13.2 基于注意力機(jī)制的 Encoder-Decoder 模型[78-80] ............................................... 238
13.3 廣義注意力機(jī)制[81-83] ......................................................................................... 240
13.4 多頭注意力機(jī)制(Multi-Head Attention)[84-87] ............................................... 241
13.5 自注意力機(jī)制(Self-Attention)[88-91] .............................................................. 242
13.6 Encoder-Decoder 架構(gòu)的變體及訓(xùn)練方法 ........................................................ 245
13.7 本章小結(jié)............................................................................................................. 249
第 14 章\tTransformer[74,94-97] .................................................................................... 250
14.1 模型介紹............................................................................................................. 251
14.2 本章小結(jié)............................................................................................................. 254
第 15 章\t深度神經(jīng)網(wǎng)絡(luò)訓(xùn)練技巧............................................................................. 255
15.1 Batch Normalization(BN) .............................................................................. 255
15.2 Layer Normalization(LN)[99] .......................................................................... 262
15.3 本章小結(jié)............................................................................................................. 263
第 16 章\t矩陣分解(Matrix Factorization)[100] ................................................ 264
16.1 理論介紹............................................................................................................. 264
16.2 本章小結(jié)............................................................................................................. 267
后記 ....................................................................................................................................... 268
參考文獻(xiàn).............................................................................................................................. 270

本目錄推薦

掃描二維碼
Copyright ? 讀書網(wǎng) m.ranfinancial.com 2005-2020, All Rights Reserved.
鄂ICP備15019699號(hào) 鄂公網(wǎng)安備 42010302001612號(hào)