Conference PaperPDF Available

Machine Learning Neural Networks Construction and Analysis in Vectorized Design Drawings

Authors:

Abstract

Machine Learning, a recently prevalent research domain in data prediction and analysis, has been widely used in a variety of fields. In the design field, especially for architectural design, a machine learning method to learn and generate design data as pixelized images has been developed in previous researches. However, proceeding pixelized image data will cause the problems of precision loss and calculation waste, since the geometric architectural design data is efficiently stored and presented as vectorized CAD files. Thus, in this article, the author developed a specific machine learning neural network to learn and predict design drawings as vectorized data, speeding up the learning and predicting process, while improving the accuracy. First, two necessary geometric tests have been successfully done, which shows the central concept of neural network construct. Then, a design rule prediction model was built to demonstrate the methods to optimize the neural network and data structure. Lastly, a generation model based on human-made design data was constructed, which can be used to predict and generate the bedroom furniture positions by inputting the boundary data of the room, door, and window.
0$&+,1( /($51,1* 1(85$/ 1(7:25.6 &216758&7,21
$1' $1$/<6,6 ,1 9(&725,=(' '(6,*1 '5$:,1*6
+$2 =+(1*DQG <8( 5(1
8QLYHUVLW\ RI 3HQQV\OYDQLD 3KLODGHOSKLD 86$
]KKDR#GHVLJQXSHQQHGX
8QLYHUVLW\ &ROOHJH /RQGRQ /RQGRQ 8.
\XHUHQ#XFODFXN
$EVWUDFW 0DFKLQH /HDUQLQJ D UHFHQWO\ SUHYDOHQW UHVHDUFK GRPDLQ
LQ GDWD SUHGLFWLRQ DQG DQDO\VLV KDV EHHQ ZLGHO\ XVHG LQ D YDULHW\ RI
ILHOGV ,Q WKH GHVLJQ ILHOG HVSHFLDOO\ IRU DUFKLWHFWXUDO GHVLJQ D PDFKLQH
OHDUQLQJ PHWKRG WR OHDUQ DQG JHQHUDWH GHVLJQ GDWD DV SL[HOL]HG LPDJHV
KDV EHHQ GHYHORSHG LQ SUHYLRXV UHVHDUFKHV +RZHYHU SURFHHGLQJ
SL[HOL]HG LPDJH GDWD ZLOO FDXVH WKH SUREOHPV RI SUHFLVLRQ ORVV DQG
FDOFXODWLRQ ZDVWH VLQFH WKH JHRPHWULF DUFKLWHFWXUDO GHVLJQ GDWD LV
HIILFLHQWO\ VWRUHG DQG SUHVHQWHG DV YHFWRUL]HG &$' ILOHV 7KXV LQ WKLV
DUWLFOH WKH DXWKRU GHYHORSHG D VSHFLILF PDFKLQH OHDUQLQJ QHXUDO QHWZRUN
WR OHDUQ DQG SUHGLFW GHVLJQ GUDZLQJV DV YHFWRUL]HG GDWD VSHHGLQJ XS
WKH OHDUQLQJ DQG SUHGLFWLQJ SURFHVV ZKLOH LPSURYLQJ WKH DFFXUDF\
)LUVW WZR QHFHVVDU\ JHRPHWULF WHVWV KDYH EHHQ VXFFHVVIXOO\ GRQH ZKLFK
VKRZV WKH FHQWUDO FRQFHSW RI QHXUDO QHWZRUN FRQVWUXFW 7KHQ D GHVLJQ
UXOH SUHGLFWLRQ PRGHO ZDV EXLOW WR GHPRQVWUDWH WKH PHWKRGV WR RSWLPL]H
WKH QHXUDO QHWZRUN DQG GDWD VWUXFWXUH /DVWO\ D JHQHUDWLRQ PRGHO EDVHG
RQ KXPDQPDGH GHVLJQ GDWD ZDV FRQVWUXFWHG ZKLFK FDQ EH XVHG WR
SUHGLFW DQG JHQHUDWH WKH EHGURRP IXUQLWXUH SRVLWLRQV E\ LQSXWWLQJ WKH
ERXQGDU\ GDWD RI WKH URRP GRRU DQG ZLQGRZ
.H\ZRUGV 0DFKLQH /HDUQLQJ $UWLILFLDO ,QWHOOLJHQFH *HQHUDWLYH
'HVLJQ *HRPHWULF 'HVLJQ
 ,QWURGXFWLRQ
 0$&+,1( /($51,1* $/*25,7+06
7KH QDPH 0DFKLQH /HDUQLQJ ZDV ILUVW LQWURGXFHG E\ 6DPXHO  WR UHSUHVHQW
D FRPSXWDWLRQDO OHDUQLQJ WKHRU\ LQ $UWLILFLDO ,QWHOOLJHQFH EXW ZLWKRXW EHLQJ
H[SOLFLWO\ SURJUDPPHG +RZHYHU WKH FRQFHSW RI 0DFKLQH /HDUQLQJ D
FRPSXWDWLRQDO PRGHO IRU QHXUDO QHWZRUNV ZDV ILUVWO\ LQYHQWHG E\ 0F&XOORFK DQG
3LWWV  EDVHG RQ WKUHVKROG ORJLF D PDWKHPDWLFDO DOJRULWKP $FFRUGLQJ WR WKH
PRGHO WKH DFWLYDWLRQ VWDWXV  RU  RI QHXURQ LQ WKH FXUUHQW OD\HU LV FDOFXODWHG
WKURXJK WKDW LQ WKH SUHYLRXV OD\HU EDVHG RQ WKH ORJLF RSHUDWLRQ UXOHV 7KXV E\
LQSXWWLQJ DQ LQLWLDO VWDWXV RI QHXURQV LQ WKH ILUVW OD\HU WKH FRPSXWDWLRQ JUDSK ZLOO
IHHGEDFN RQ DQ RXWSXW VWDWXV RI QHXURQV LQ WKH ODVW OD\HU 7KLV WKHRU\ QRW RQO\
5( $QWKURSRFHQH 3URFHHGLQJV RI WKH WK ,QWHUQDWLRQDO &RQIHUHQFH RI WKH $VVRFLDWLRQ IRU &RPSXWHU$LGHG
$UFKLWHFWXUDO 'HVLJQ 5HVHDUFK LQ $VLD &$$'5,$  9ROXPH     DQG SXEOLVKHG E\ WKH
$VVRFLDWLRQ IRU &RPSXWHU$LGHG $UFKLWHFWXUDO 'HVLJQ 5HVHDUFK LQ $VLD &$$'5,$ +RQJ .RQJ
 + =+(1* $1' < 5(1
HVWDEOLVKHG WKH IRXQGDWLRQV LQ WKH ELRORJLF SURFHVVHV RI WKH EUDLQ EXW DOVR LQVSLUHG
WKH QHXUDO QHWZRUN VWUXFWXUH RI $UWLILFLDO ,QWHOOLJHQFH
7KHQ :HUERV  GHYHORSHG D PHWKRG FDOO %DFN 3URSDJDWLRQ WR DFFHOHUDWH
WKH WUDLQLQJ RI PXOWLOD\HU QHXUDO QHWZRUNV ZKLFK IHHGV WKH HUURU WHUP EDFN WR
WKH QHXUDO QHWZRUN WR PRGLI\ WKH SDUDPHWHUV LQ HDFK QHXURQ %\ FDOFXODWLQJ WKH
JUDGLHQW RI WKH ORVV IXQFWLRQ EHWZHHQ WKH JURXQG WUXWK YDOXH DQG WKH SUHGLFWHG
YDOXH VXFK DV PHDQ VTXDUH HUURU WKH SDUDPHWHUV LQ WKH KLGGHQ OD\HUV ZLOO EH
XSGDWHG DQG JUDGXDOO\ DSSURDFK WKH RSWLPDO YDOXHV :LWK HQRXJK GDWD DQG
SURFHVVLQJ SRZHU RI FRPSXWHUV WKH ZHLJKWV RI HDFK QHXURQ LQ WKH QHXUDO QHWZRUN
ZLOO ILQDOO\ UHDFK D FRQGLWLRQ WKDW WKH WRWDO HUURU YDOXH ORVV IXQFWLRQ LV PLQLPL]HG
7KH QHXUDO VWUXFWXUH DQG WKH EDFNSURSDJDWLRQ DOJRULWKP WRJHWKHU IRUP WKH EDVLF
XQLW RI WKH QHXUDO QHWZRUN ZKLFK IXUWKHU LQVSLUHG WKH GHYHORSPHQW RI 5HFXUUHQW
1HXUDO 1HWZRUNV 511 UHSHDWHGO\ DSSO\LQJ WKH EDVLF XQLW WR SURFHHG VHTXHQWLDO
GDWD ODQJXDJH PRGHO DQG &RQYROXWLRQDO 1HXUDO 1HWZRUNV &11 H[WHQGLQJ WKH
GLPHQVLRQ RI WKH EDVLF XQLW WR SURFHHG PDWUL[ GDWD LPDJH PRGHO ILJXUH 
)LJXUH  1HXUDO 1HWZRUN 6WUXFWXUHV RI 'LIIHUHQW 0DFKLQH /HDUQLQJ $OJRULWKPV
 35(9,286 :25. ,1 '(6,*1 '5$:,1*6
,PDJHV HVSHFLDOO\ GUDZLQJV DV WKH UHSUHVHQWDWLYH RI GHVLJQ KDYH EHHQ FRPPRQO\
XVHG WR VWRUH DQG SUHVHQW GHVLJQ GDWD +XDQJ DQG =KHQJ  DSSOLHG D
*HQHUDWLYH $GYHUVDULDO 1HWZRUN *$1 D UHILQHG YHUVLRQ RI &11 WR WUDLQ DQG
SUHGLFW DUFKLWHFWXUDO LPDJH GDWD LQ SDLUV PDSSLQJ WKH SDUDPHWHUV EHWZHHQWZR
LPDJHV +RZHYHU WKHUH DUH VWLOO EOXUU\ DQG XQFOHDU DUHDV DQG WKH IXUQLWXUH LV
QRW UHFRJQL]DEOH DV LW VKRXOG EH :KDW¶V PRUH WKH PRVW VLJQLILFDQW OLPLWDWLRQ
RI &11 LV WKDW LW UHJDUGV GHVLJQ GDWD DV SL[HOL]HG LPDJHV +HQFH WKH SUHFLVLRQ
LV OLPLWHG DQG LW GHSHQGV RQ WKH QXPEHU RI SL[HOV LQ WKH LPDJH (QODUJLQJ WKH
VL]H RI WKH LPDJH FDQ LPSURYH LWV SHUIRUPDQFH EXW ZLOO DOVR LQFUHDVH WKH WUDLQLQJ
WLPH 6LPLODU UHVHDUFK RI DSSO\LQJ LPDJHEDVHG QHXUDO QHWZRUNV LQ JHQHUDWLYH
WDVNV LQFOXGHV .LQXJDZD DQG 7DNL]DZD  1HZWRQ  6WHLQIHOG 3DUNHW
DO  7KRPVHQ 1LFKRODV HW DO  7XUORFN DQG 6WHLQIHOG  =DQGDYDOL
DQG *DUFtD 
+RZHYHU =KHQJ DQG +XDQJ  DOVR SURSRVHG D PHWKRG LQ WKHLU SDSHU
WR YLVXDOL]H WKH SDUDPHWHUV LQ WKH QHXUDO QHWZRUN 7KH FRQYROXWLRQ OD\HUV LQ
&11 ZRUN WR GHWHFW WKH ERXQGDU\ LQIRUPDWLRQ IURP WKH RULJLQDO LPDJH DQG WKHQ
0$&+,1( /($51,1* 1(85$/ 1(7:25.6 &216758&7,21
$1' $1$/<6,6 ,1 9(&725,=(' '(6,*1 '5$:,1*6

FRPELQH IHDWXUHV LQWR D FKDRWLF V\VWHP IRU FDOFXODWLRQ
7KXV &11 ILUVWO\ WUDQVODWHV LPDJH GDWD LQWR YHFWRUOLNH GDWD EXW VWLOO VWRUHG
DV D PDWUL[ E\ FUHDWLQJ FRQWLQXRXV ZKLWH RU EODFN SL[HOV WKHQ OLQNV WKH SL[HO
GDWD WR HVVHQWLDO QHXUDO QHWZRUNV IRU RXWSXWWLQJ D VLQJOH YDOXH LPDJH FODVVLILHU RU
WR GHFRQYROXWLRQ OD\HUV IRU RXWSXWWLQJ DQRWKHU LPDJH LPDJH JHQHUDWRU $ ODUJH
DPRXQW RI SURFHVVLQJ SRZHU LV XVHG WR XSGDWH PLOOLRQV RI SDUDPHWHUV ZKLFK RQO\
DFW WR WUDQVIRUP LPDJHV LQWR YHFWRUL]HG GDWD 7KXV LW¶V XQQHFHVVDU\ WR WUDLQ &11V
LI YHFWRUL]HG GDWD DOUHDG\ H[LVWV LQ &$' ILOHV LQ PRVW RI WKH DUFKLWHFWXUDO GHVLJQ
FDVHV
 0HWKRGRORJ\
 %$6,& *(20(75,& 7(67  ,16&5,%(' &,5&/(
,QVSLUHG E\ WKH IDFWV DERYH WZR WHVWV LQ FRQVWUXFWLQJ QHXUDO QHWZRUNV WR WUDLQ
VLPSOH JHRPHWULF GDWD ZHUH FDUULHG RXW ILUVWO\ YDOLGDWLQJ WKH IHDVLELOLW\ RI WUDLQLQJ
DQG SUHGLFWLQJ YHFWRUL]HG GDWD
)LUVW D VDPSOH QHXUDO QHWZRUN ZDV EXLOW WR SUHGLFW WKH LQVFULEHG FLUFOH E\
LQSXWWLQJ D UHFWDQJOH $V ILJXUH  VKRZV WKH LQSXWWHG VTXDUH LV UHSUHVHQWHG E\
WKH FRRUGLQDWHV RI LWV WZR YHUWH[HV DV (x1,y1) DQG (x2,y2) 8VLQJ WKH VDPH UXOH
LWV LQVFULEHG FLUFOH LV UHSUHVHQWHG E\ WKH FRRUGLQDWH RI WKH FHQWHU SRLQW (x3,y3) DQG
WKH UDGLXV r 7KHUHIRUH WKH ILUVW OD\HU LQ WKH QHXUDO QHWZRUN VKRXOG FRQWDLQ IRXU
QHXURQV DQG WKH ODVW OD\HU VKRXOG FRQWDLQ WKUHH QHXURQV
)LJXUH  *HRPHWULF 6WDWHPHQW DQG 1HWZRUN 6WUXFWXUH RI ,QVFULEHG &LUFOH 3UHGLFWLRQ
%HIRUH GHILQLQJ KRZ PDQ\ KLGGHQ OD\HUV WKH QHXUDO QHWZRUN VKRXOG KDYH
PDWKHPDWLFDOO\ WKH YDOXH RI x1y1x2y2x3y3DQGrFDQ EH H[SUHVVHG
DV D OLQHDU IRUPXOD &RPSDUHG WR WKH DFWLYDWLRQ IXQFWLRQ ZKLFK GHVFULEHV WKH
FDOFXODWLRQ PHWKRG IURP WKH FXUUHQW OD\HU QHXURQV WR WKH QH[W OD\HU QHXURQV WKH
LQGH[HV RI WKH JURXQG WUXWK IRUPXOD DQG WKH DFWLYDWLRQ IXQFWLRQ IRU RQH OD\HU DUH
ERWK  7KHUH LV QR QHHG WR FKDQJH WKH DFWLYDWLRQ IXQFWLRQ WR LQFUHDVH WKH LQGH[
RU WR DGG DQ\ KLGGHQ OD\HUV WR FRPSOLFDWH WKH QHXUDO QHWZRUN D QHXUDO QHWZRUN
 + =+(1* $1' < 5(1
ZLWKRXW KLGGHQ OD\HUV VKRXOG SHUIRUP EHWWHU WR DYRLG WKH RYHUILWWLQJ SUREOHP
7KXV DIWHU WUDLQLQJ ZLWK D GDWDVHW RI  SDLUV RI UHFWDQJOHV DQG FLUFOHV
ZKLFK RQO\ WRRN  VHFRQGV WR ORRS  WLPHV ZLWKRXW *38 VXSSRUW WKH QHWZRUN
SHUIRUPHG QLFHO\ WR SUHGLFW WKH YDOXHV 7KLV FDVH GHPRQVWUDWHV WKDW UDWKHU
WKDQ GLUHFWO\ DSSO\LQJ FRPSOH[ QHXUDO QHWZRUNV WR WKH SUREOHPV DQDO\]LQJDQG
UHIHUULQJ WKH SRVVLEOH PDWKHPDWLFDO VROXWLRQ LV YHU\ LPSRUWDQW IRU VLPSOLI\LQJ WKH
QHXUDO QHWZRUNV DQG LPSURYLQJ LWV HIILFLHQF\
 $'9$1&(' *(20(75,& 7(67  63/,1( ,17(532/$7,21
1H[W D QHXUDO QHWZRUN WR SUHGLFW WKH FXELF VSOLQH LQWHUSRODWLRQ E\ LQSXWWLQJ WKUHH
FRQWURO SRLQWV ZDV EXLOW $V ILJXUH  VKRZV WR VLPSOLI\ WKH SUREOHP D VSOLQH LV
GLYLGHG LQWR  SRLQWV ZKLFK DSSUR[LPDWHO\ UHSUHVHQWV WKH FXUYH 7KHUHIRUH WKH
ILUVW OD\HU RI WKH QHXUDO QHWZRUN VKRXOG EH WKH FRRUGLQDWHV RI WKH WKUHH FRQWURO
SRLQWV ZLWK VL[ LQSXW QHXURQV ZKLOH WKH ODVW OD\HU VKRXOG EH WKH FRRUGLQDWHV RI WKH
 GLYLGHG SRLQWV ZLWK  RXWSXW QHXURQV
)LJXUH  *HRPHWULF 6WDWHPHQW DQG 1HWZRUN 6WUXFWXUH RI 6SOLQH ,QWHUSRODWLRQ 3UHGLFWLRQ
%HIRUH FRQVWUXFWLQJ WKH KLGGHQ OD\HU WKH PDWKHPDWLFDO H[SUHVVLRQ RI FXELF
VSOLQH LQWHUSRODWLRQ LQ WKLV FDVH LV WKH IRUPXOD DV IROORZLQJ
yi=aL ·(xix1)3+bL ·(xix1)2+cL ·(xix1)1+dL 
*HQHUDOO\ VSHDNLQJ WR FDOFXODWH WKH RXWSXW QHXURQ YDOXHV x(i)DQG y(i) DIWHU
SDVVLQJ WKURXJK WKH QHXUDO QHWZRUN WKH LQGH[ RI WKH LQSXW QHXURQ YDOXHV VKRXOG
EH D FRPELQDWLRQ RI   DQG  ZKLFK PHDQV WKH DFWLYDWLRQ IXQFWLRQ VKRXOG QRW
VWLOO EH OLQHDU EXW D SRO\QRPLDO IXQFWLRQ DV
y1=w1·x3
1+w2·x2
1+w3·x1
1+b1
$OVR IRU HYHU\ VSOLQH WKH QHXUDO QHWZRUN SDUDPHWHUV DUH GLIIHUHQW 7KXV WR FRYHU
DOO VLWXDWLRQV DV IDU DV SRVVLEOH D KLGGHQ OD\HU ZLWK  QHXURQV LV QHHGHG VR WKDW
WKH FRPELQDWLRQ RI SDUDPHWHUV ZLOO EH HQODUJHG WR PDWFK PRUH GDWD 7KH DFWLYDWLRQ
IXQFWLRQ IRU WKH KLGGHQ OD\HU LV WKH VLJPRLG IXQFWLRQ LQFUHDVLQJ WKH JUDGLHQW DQG
VSHHGLQJ XS WKH WUDLQLQJ SURFHVV
$IWHU WUDLQLQJ WKLV QHXUDO QHWZRUN UHDFKHG YHU\ KLJK DFFXUDF\ 7KXV LQ
FRQFOXVLRQ WKH QHXUDO QHWZRUN WR SUHGLFW VSOLQH LQWHUSRODWLRQ FRQWDLQV RQH LQSXW
OD\HU RQH KLGGHQ OD\HU DQG RQH RXWSXW OD\HU EXW ZKDW¶V LPSRUWDQW LV WKDW WKH
0$&+,1( /($51,1* 1(85$/ 1(7:25.6 &216758&7,21
$1' $1$/<6,6 ,1 9(&725,=(' '(6,*1 '5$:,1*6

DFWLYDWLRQ IXQFWLRQ LV FXVWRPL]HG WR VROYH WKLV SUREOHP HVSHFLDOO\ 7KLV FDVH
GHPRQVWUDWHV WKDW IXQFWLRQV LQ WKH QHXUDO QHWZRUNV VXFK DV WKH DFWLYDWLRQ IXQFWLRQ
DQG WKH ORVV IXQFWLRQ VKRXOG DOVR EH FRQVLGHUHG DQG FXVWRPL]HG WR RSWLPL]HWKH
QHXUDO QHWZRUNV
 '(6,*1 58/( 7(67  5(&7$1*/( 3$77(51
2WKHU WKDQ GLUHFWO\ FUHDWLQJ VLPSOH JHRPHWULF GDWD OLNH LQVFULEHG FLUFOH RU VSOLQH
LQWHUSRODWLRQ ILJXUH  VKRZV D SDWWHUQ GHVLJQ WKDW IRU D JLYHQ ERXQGDU\ ZLWK WKH
YHUWH[HV RI (x1,y1)(x2,y2)(x3,y3)DQG(x4,y4) UHFWDQJOHV ZLWK VLGH OHQJWKV
RI r1DQG r2VKRXOG EH ILWWHG LQWR WKH ERXQGDU\ DV PDQ\ DV SRVVLEOH 6WLOO D JDS ZLWK
WKH OHQJWK RI  XQLW VKRXOG UHPDLQ EHWZHHQ UHFWDQJOHV DQG WKH FHQWHU SRLQWV RI
WKH UHFWDQJOHV DUH PDUNHG DV (x11,y11)  (x14,y14)(x21,y21)  (x24,y24)
DQG VR RQ E\ WKH QXPEHUV RI URZV DQG FROXPQV
)LJXUH  *HRPHWULF 6WDWHPHQW DQG 1HWZRUN 6WUXFWXUH RI 6SOLQH 5HFWDQJOH 3DWWHUQ
$FFRUGLQJ WR WKH YDULDEOHV GHVFULEHG DERYH D QHXUDO QHWZRUN ZLWK WKH LQSXW
OD\HU RI  QHXURQV ZKLFK UHSUHVHQW WKH FRRUGLQDWHV RI WKH IRXU YHUWH[HV DQG WKH
YDOXHV RI r1DQG r2 ZDV SURSRVHG DW ILUVW %XW WKH SUREOHP LV WKH QXPEHU RI
QHXURQV LQ WKH RXWSXW OD\HU $V ZH NQRZ WKH WRWDO QXPEHU RI UHFWDQJOHV WKDW FDQ
EH ILWWHG LQWR WKH ERXQGDU\ LV GLIIHUHQW IRU HYHU\ LQSXW GDWD %XW WKH QHXUDO QHWZRUN
PRGHO UHTXLUHV D FHUWDLQ DPRXQW RI QHXURQV 6R WKH VROXWLRQ LV WKDW DQ RXWSXW OD\HU
ZLWK  QHXURQV ZDV FRQVWUXFWHG LQ WKH QHXUDO QHWZRUN WR UHSUHVHQW D PD[LPXP
RI  FHQWHU SRLQWV 2WKHU WKDQ DVVLJQLQJ WKH FRRUGLQDWHV RI WKH H[LVWLQJ SRLQWV WR
WKH QHXURQV WKH YDOXHV RI WKH UHVW QHXURQV ZLOO EH  WR LQGLFDWH WKH LQYDOLG SRLQWV
$OVR LQVWHDG RI VRUWLQJ YDOLG YDOXHV WRJHWKHU WKH RXWSXW OD\HU DUUDQJHV WKH GDWD
E\ LWV URZ DQG FROXPQV (YHU\  QHXURQV UHSUHVHQW WKH UHFWDQJOHV LQ RQH URZVR
RQO\ WKH QHXURQV IRU WKH UHFWDQJOHV LQ WKH VDPH URZ DQG FROXPQ ZLOO VKDUH WKH VDPH
SDUDPHWHUV 7KHQ WKH LQYDOLG YDOXH RI µ¶ ZLOO QRW LQIOXHQFH WKH WUDLQLQJ SURFHVV
7KH ODWHU VXFFHVVIXO WUDLQLQJ RI WKH QHXUDO QHWZRUN LQ WKLV FDVH SUHVHQWVD
SURVSHFW WKDW ZLWK VXLWDEOH QHWZRUN IUDPH GDWD VWUXFWXUH DFWLYDWLRQ IXQFWLRQ DQG
ORVV IXQFWLRQ GHVLJQ DV ORQJ DV LW LV GDWD KDV WKH SRWHQWLDO WR EH OHDUQHG DQG
 + =+(1* $1' < 5(1
SUHGLFWHG WKURXJK 0DFKLQH /HDUQLQJ
 $SSOLFDWLRQ
 /($51,1* $1' *(1(5$7,1* 0$10$'( '(6,*1 '$7$
3UHYLRXV WHVWV DUH DOO EDVHG RQ FOHDU GHVLJQ UXOHV DQG DXWRJHQHUDWHG GDWDVHWV
:LWK WKH VXFFHVVIXO H[SHULPHQWV D VLPLODU QHXUDO QHWZRUN ZDV EXLOW WR OHDUQ DQG
SUHGLFW KXPDQ GHVLJQ GDWD 7KH QHXUDO QHWZRUN SDUDPHWHUV ZHUH DOVR DQDO\]HG WR
VXPPDUL]H WKH SRVVLEOH IRUPXODV RI KXPDQPDGH GHVLJQ
,Q WKH WHVW RI WKH UHFWDQJOH SDWWHUQ WKH QHXUDO QHWZRUN PRGHO FDQ SUHGLFW '
JHRPHWULHV 7KHUHIRUH D QHXUDO QHWZRUN ZLWK VLPLODU VWUXFWXUH DQG IXQFWLRQV
VKRXOG ZRUN ZKHQ EHLQJ DSSOLHG WR WKH ' DUFKLWHFWXUDO GUDZLQJV ZKLFK LQ IDFW
DUH WKH SULPDU\ FRPPXQLFDWLRQ WRROV EHWZHHQ DUFKLWHFWV
7R VLPSOLI\ WKH TXHVWLRQ D GDWDVHW RI IORRU SODQV RI EHGURRP GHVLJQ ZDV
FROOHFWHG IURP OLDQMLDFRP IRU WUDLQLQJ DQG WHVWLQJ )LUVW WKH HOHPHQWV LQ WKH
EHGURRP GHVLJQ ZHUH PDUNHG DV ILJXUH  VKRZV 5HG IRU WKH EHGURRP ERXQGDU\
F\DQ IRU WKH GRRU SXUSOH IRU WKH ZLQGRZ EOXH IRU WKH EHG JUHHQ IRU EHG VWDQG
DQG RUDQJH IRU WKH 79 VHW (YHU\ PDUN LV D UHFWDQJOH UHSUHVHQWLQJ D VSHFLILFDUHD
)LJXUH  /DEHOOLQJ 5XOHV RI %HGURRP 3ODQ 'UDZLQJV
7KH QHXUDO QHWZRUN VKRXOG WDNH LQ WKH LQIRUPDWLRQ RI EHGURRP ERXQGDU\ GRRU
DQG ZLQGRZ DV WKH URRP¶V VLWXDWLRQ XQGHU GHVLJQ WKHQ RXWSXW WKH UHFWDQJOHV RI
EHG EHG VWDQGV LI H[LVWLQJ DQG 79 VHW LI H[LVWLQJ DV WKH GHVLJQ GDWD 7KXV
WKHUH VKRXOG EH WKUHH UHFWDQJOHV LQ WKH LQSXW OD\HU DQG  WR  UHFWDQJOHV LQ WKH
RXWSXW OD\HU
 1(85$/ 1(7:25. 6758&785(
7R SUHSDUH WKH GDWD IRU WUDLQLQJ DV ILJXUH  VKRZV WKH JHRPHWULHV ZHUH ILUVW PRYHG
WR WKH RULJLQDO SRLQW EDVHG RQ WKH OHIWGRZQ YHUWH[ RI WKH EHGURRP ERXQGDU\ VR WKDW
WKH EHGURRP ERXQGDU\ FDQ EH UHSUHVHQWHG DV (x1,y1) 7KHQ IRU WKH GRRU DQG WKH
ZLQGRZ(x2,y2) DQG (x3,y3) UHSUHVHQW WKHLU FHQWHU SRLQWV ZKLOH r21r22r31
DQG r32 UHSUHVHQW WKHLU VLGH OHQJWKV 7KHUHIRUH WKH LQSXW OD\HU VKRXOG FRQWDLQ 
QHXURQV (x1,y1,x2,y2,r21,r22,x3,y3,r31,r32)
0$&+,1( /($51,1* 1(85$/ 1(7:25.6 &216758&7,21
$1' $1$/<6,6 ,1 9(&725,=(' '(6,*1 '5$:,1*6

)LJXUH  9DULDEOHV DQG 0HDQLQJV RI %HGURRP 3ODQ *HQHUDWLRQ
)RU WKH RXWSXW OD\HU LW¶V XQFHUWDLQ KRZ PDQ\ LWHPV VKRXOG EH FRQWDLQHG LQ WKH
URRP 7KHUHIRUH LQ DGGLWLRQ WR XVLQJ WKH VDPH UXOH WR UHFRUG WKH UHFWDQJOH DV WKH
FHQWHU SRLQW DQG VLGH OHQJWKV D YDULDEOH eLV LQWURGXFHG DV WKH ILUVW QHXURQ ZKLFK
UHSUHVHQWV WKH SRVVLELOLW\ WKDW WKH UHFWDQJOH EHORZ H[LVWV 7KDW PHDQV LQWKH
WUDLQLQJ GDWDVHW D GDWD RI (0,1,1,1,1) PHDQV WKHUH LV QR UHFWDQJOHV D GDWD
RI (1,0.5,0.5,0.2,0.2) PHDQV WKHUH LV D UHFWDQJOH ZKRVH FHQWULF SRLQW LV (0.5,0.5)
DQG VLGH OHQJWKV DUH ERWK  7KXV LI D WHVWLQJ UHVXOW VKRZV DQ eYDOXH VPDOOHU
WKDQ  ZH FDQ DVVXPH WKDW WKHUH VKRXOG QRW EH DQ LWHP DQG ZH VKRXOG LJQRUH
WKH YDOXHV LQ WKH IROORZLQJ IRXU QHXURQV $IWHU DUUDQJLQJ WKH RXWSXW  QHXURQV DV
(e4,x4,y4,r41,r42,e5,x5,y5,r51,r52,e6,x6,y6,r61,r62,e7,x7,y7,r71,r72)
WKH QHXUDO QHWZRUN VWUXFWXUH LV VKRZQ LQ ILJXUH 
)LJXUH  1HXUDO 1HWZRUN 6WUXFWXUH RI %HGURRP 3ODQ *HQHUDWLRQ
$FFRUGLQJ WR WKH SUHYLRXV WHVW WKH DFWLYDWLRQ IXQFWLRQ DQG WKH ORVV IXQFWLRQ RI
WKH QHXUDO QHWZRUN VKRXOG EH WKH VDPH DV WKRVH RI WKH UHFWDQJOH SDWWHUQ
y=w·x+b
LOSS(y, ˆy)= 1
n
n
i=1
(yiyi))2y0
LOSS(y, ˆy)=0 y<0
 + =+(1* $1' < 5(1
 5(68/7 $1$/<6,6
:LWK D GDWDVHW RI  EHGURRP SODQ GUDZLQJV  RI WKHP ZHUH FKRVHQ DV WKH
WUDLQLQJ VHW DQG WKH UHPDLQLQJ  RI WKHP ZHUH WHVWHG DV WKH HYDOXDWLQJ VHW7KH
WUDLQLQJ SURFHVV RQO\ WRRN  VHFRQGV WR ORRS  WLPHV ZLWKRXW *38 VXSSRUW
VLQFH WKH QHXUDO QHWZRUN LV TXLWH VLPSOH DQG WKH WUDLQLQJ GDWD LV WLQ\
)LJXUH  VKRZV WKH VHOHFWHG JHQHUDWLRQ UHVXOWV ZKLFK LQGLFDWHV D IDFW WKDWWKH
SUHGLFWHG GHVLJQV DUH QRW DOZD\V WKH VDPH DV WKH JURXQG WUXWK GHVLJQV 7R EH
VSHFLILF LQ QR QR DQG QR ZKLOH WKH SUHGLFWHG SRVLWLRQV RI WKH EHG
DQG EHG VWDQGV DUH KLJKO\ VLPLODU WR WKH JURXQG WUXWK GHVLJQ WKH QHXUDO QHWZRUN
JHQHUDWHG DQ DGGLWLRQDO 79 VHW ZKLFK GRHV QRW DSSHDU LQ WKH JURXQG WUXWK GHVLJQ
EXW LV UHDVRQDEOH ,Q QR DQG QR WKH PRGHO DOVR JHQHUDWHG D 79 VHW EXW
WKH SRVLWLRQ LV VOLJKWO\ GLIIHUHQW ZKLFK LV FORVHU WR WKH FHQWHU D[LV RI WKH URRP
)RU VPDOO URRPV OLNH QR QR DQG QR WKHUH DUH QR 79 VHWV ERWK LQ WKH
JHQHUDWHG DQG WKH JURXQG WUXWK GHVLJQV
$V IRU WKH EHG DQG WKH EHG VWDQGV LQ WKH WUDLQLQJ VHW WKHUH LV DOZD\V D EHG
LQ D EHGURRP VR LW¶V QR GRXEW WKDW WKHUH VKRXOG DOVR EH D EHG LQ HYHU\ URRP LI
WKH LQSXW GDWD LQ WKH WHVWLQJ VHW LV UHDVRQDEOH +RZHYHU WKHUH DUH HLWKHU WZR EHG
VWDQGV ZKLFK QHLJKERU ZLWK WKH EHG RU QR EHG VWDQGV LQ WKH WUDLQLQJ VHW VR WKH
YDULDEOH µH¶ DFWV DV WKH PDLQ UROH WR WHOO ZKHWKHU WKHUH VKRXOG EH EHG VWDQGV RU QRW
,Q QR UDWKHU WKDQ JHQHUDWLQJ WZR EHG VWDQGV WKH QHXUDO QHWZRUN SUHIHUV WR
RXWSXW D PRUH PDVVLYH EHG ZLWKRXW EHG VWDQGV $OVR LQ QR WKH EHG VWDQGV
GLVDSSHDU IURP WKH JURXQG WUXWK GDWD WR WKH JHQHUDWHG GDWD ZKLFK PLJKW EH FDXVHG
E\ WKH VKRUW VLGH OHQJWK RI WKH URRP )XUWKHU WKH SRVLWLRQ RI WKH EHG DQG WKHEHG
VWDQGV DUH FORVHU WR WKH FHQWHU D[LV IRU H[DPSOH LQ WKH VLWXDWLRQ IRU WKH 79 VHW
HVSHFLDOO\ LQ QR WKHUH LV D PDVVLYH VKLIW EHWZHHQ WKH JURXQG WUXWK GDWD DQG WKH
JHQHUDWHG GDWD +RZHYHU WKH JHQHUDWHG UHVXOWV DUH PRUH UHDVRQDEOH $QG IURP WKH
PDWKHPDWLFDO DVSHFW WKH SDUDPHWHUV LQ WKH QHXUDO QHWZRUN UHSUHVHQW WKH ZHLJKWHG
UHVXOWV RI DOO WKH GDWD LQ WKH WUDLQLQJ VHW 7KXV WKH QHXUDO QHWZRUN VKRXOG DOZD\V
RXWSXW WKH PRVW RSWLPL]HG VROXWLRQ EDVHG RQ WKH WUDLQLQJ GDWD
0$&+,1( /($51,1* 1(85$/ 1(7:25.6 &216758&7,21
$1' $1$/<6,6 ,1 9(&725,=(' '(6,*1 '5$:,1*6

)LJXUH  3HUIRUPDQFH &\DQ IRU ,QSXW %OXH IRU *URXQG 7UXWK 5HG IRU *HQHUDWLRQ
)LJXUH  VKRZV WKH FRPSDULVRQ RI WKH UHVXOWV EHWZHHQ WKH LPDJHEDVHG QHXUDO
QHWZRUN DQG RXU YHFWRUEDVHG QHXUDO QHWZRUN )RU WKH XSSHU EHGURRP ERWK WZR
QHXUDO QHWZRUNV JHQHUDWHG WZR EHG VWDQGV ZKLFK LV GLIIHUHQW IURP WKH RULJLQDO
GHVLJQ +RZHYHU IRU WKH ORZHU EHGURRP WKH LPDJHEDVHG QHXUDO QHWZRUN
RXWSXWWHG D QRLVH LPDJH ZKLOH RXU YHFWRUEDVHG QHXUDO QHWZRUN IHG EDFN D
UHFWDQJOH VKRZLQJ WKH SRVLWLRQ DQG VL]H RI WKH EHG 7KLV UHVXOW GHPRQVWUDWHV WKH
VWDELOLW\ RI RXU YHFWRUEDVHG QHXUDO QHWZRUN
)LJXUH  ,PDJHEDVHG 3UHGLFWLRQ DQG 9HFWRUEDVHG 3UHGLFWLRQ
 &RQFOXVLRQ
0DFKLQH /HDUQLQJ LV D SRZHUIXO WRRO IRU SURFHVVLQJ GHVLJQ GDWD HVSHFLDOO\ IRU
DUFKLWHFWXUDO GHVLJQ GUDZLQJV /HDUQLQJ DQG SUHGLFWLQJ DUFKLWHFWXUDO GUDZLQJV
DV YHFWRUL]HG GDWD LV PRUH HIILFLHQW DQG DFFXUDWH WKDQ WKDW DV LPDJH GDWD )RU
 + =+(1* $1' < 5(1
VLPSOH WDVNV VXFK DV JHQHUDWLQJ SODQ GUDZLQJV RI LQWHULRU GHVLJQ D SULPDU\ QHXUDO
QHWZRUN FDQ DFW FRUUHFWO\ WR VDWLVI\ WKH GHVLJQ UHTXLUHPHQWV ZKLFK VKRZV WKH
SRWHQWLDO RI 0DFKLQH /HDUQLQJ LQ GHVLJQ SUDFWLFH
,Q WKH IXWXUH WKH WHQGHQF\ RI WKH GHVLJQ FRRSHUDWLRQ EHWZHHQ KXPDQV DQG
PDFKLQHV ZLOO EHFRPH PRUH DSSDUHQW 7KH PDFKLQH ZLOO DVVLVW WKH GHVLJQ SURFHVV
QRW RQO\ LQ VLPSOH UHSHDWHG ZRUN EXW DOVR LQ FUHDWLYH ZRUN E\ OHDUQLQJ WKH GHVLJQ
H[DPSOHV IURP WKH KXPDQ $UFKLWHFWXUH GHVLJQ LV D YDVW WRSLF KRZHYHU E\
GLYLGLQJ WKH ZKROH GHVLJQ SURFHVV LQWR VHYHUDO VPDOO DQG VWUDLJKWIRUZDUG WDVNV DQG
XVLQJ VSHFLDOO\WUDLQHG QHXUDO QHWZRUNV WR VROYH HDFK RI WKHP GHVLJQ E\ PDFKLQH
ZLOO H[FHHG WKH KXPDQ DELOLW\ 7KHUHIRUH WKH IROORZLQJ UHVHDUFK RI WKLV SDSHU LV WR
GHYHORS VSHFLILF QHXUDO QHWZRUNV WR GHDO ZLWK GLIIHUHQW W\SHV RI GHVLJQ WDVNV DQG
ILQDOO\ EXLOG D ³PDVWHU GHVLJQHU´ E\ PDFKLQH
5HIHUHQFHV
+XDQJ : DQG =KHQJ +  $UFKLWHFWXUDO 'UDZLQJV 5HFRJQLWLRQ DQG *HQHUDWLRQ WKURXJK
0DFKLQH /HDUQLQJ 3URFHHGLQJV RI $&$',$  0H[LFR &LW\ 0H[LFR
.LQXJDZD + DQG 7DNL]DZD $  'HHS /HDUQLQJ 0RGHO IRU 3UHGLFWLQJ 3UHIHUHQFH RI 6SDFH
E\ (VWLPDWLQJ WKH 'HSWK ,QIRUPDWLRQ RI 6SDFH XVLQJ 2PQLGLUHFWLRQDO ,PDJHV 3URFHHGLQJV
RI (&$$'( 6,*5$',  3RUWR 3RUWXJDO
0F&XOORFK :6 DQG 3LWWV :  $ ORJLFDO FDOFXOXV RI WKH LGHDV LPPDQHQW LQ QHUYRXV
DFWLYLW\ 7KH EXOOHWLQ RI PDWKHPDWLFDO ELRSK\VLFV 
1HZWRQ '  'HHS *HQHUDWLYH /HDUQLQJ IRU WKH *HQHUDWLRQ DQG $QDO\VLV RI $UFKLWHFWXUDO
3ODQV ZLWK 6PDOO 'DWDVHWV 3URFHHGLQJV RI (&$$'( 6,*5$',  3RUWR 3RUWXJDO
6DPXHO $/  6RPH VWXGLHV LQ PDFKLQH OHDUQLQJ XVLQJ WKH JDPH RI FKHFNHUV ,%0 -RXUQDO
RI UHVHDUFK DQG GHYHORSPHQW 
6WHLQIHOG . 3DUN . 0HQJHV $ DQG :DONHU 6  )UHVK (\HV  $ IUDPHZRUN IRU WKH
DSSOLFDWLRQ RI PDFKLQH OHDUQLQJ WR JHQHUDWLYH DUFKLWHFWXUDO GHVLJQ DQG D UHSRUW RI DFWLYLWLHV
DW 6PDUWJHRPHWU\  3URFHHGLQJV RI &$$' )XWXUHV  'DHMHRQ .RUHD
7KRPVHQ 05 1LFKRODV 3 7DPNH 0 *DW] 6 DQG 6LQNH <  3UHGLFWLQJ DQG VWHHULQJ
SHUIRUPDQFH LQ DUFKLWHFWXUDO PDWHULDOV 3URFHHGLQJV RI (&$$'( 6,*5$',  3RUWR
3RUWXJDO
7XUORFN 0 DQG 6WHLQIHOG .  1HFHVVDU\ 7HQVLRQ $ 'XDO(YDOXDWLRQ *HQHUDWLYH 'HVLJQ
0HWKRG IRU 7HQVLRQ 1HW 6WUXFWXUHV 3URFHHGLQJV RI WKH 'HVLJQ 0RGHOOLQJ 6\PSRVLXP 
%HUOLQ *HUPDQ\
:HUERV 3  %H\RQG UHJUHVVLRQ QHZ IRROV IRU SUHGLFWLRQ DQG DQDO\VLV LQ WKH EHKDYLRUDO
VFLHQFHV 3K' 7KHVLV +DUYDUG 8QLYHUVLW\
=DQGDYDOL %$ DQG *DUFtD 0-  $XWRPDWHG %ULFN 3DWWHUQ *HQHUDWRU IRU 5RERWLF
$VVHPEO\ XVLQJ 0DFKLQH /HDUQLQJ DQG ,PDJHV 3URFHHGLQJV RI (&$$'( 6,*5$', 
3RUWR 3RUWXJDO
=KHQJ + DQG +XDQJ :  8QGHUVWDQGLQJ DQG 9LVXDOL]LQJ *HQHUDWLYH $GYHUVDULDO
1HWZRUN LQ $UFKLWHFWXUDO 'UDZLQJV 3URFHHGLQJV RI &$$'5,$  %HLMLQJ &KLQD
... Cho et al. [8] used the floor plan of a building as input data and the semantic segmentation results of various labeled elements as the output to obtain a GAN that could identify and label the elements in a building floor plan. Further studies, such as Zheng et al. [9], used vectorized coordinate values instead of pixel maps for the deep learning of a floor plan layout. The design elements in the house plan were vectorized into coordinate values to study a bedroom layout problem in the floor plan, and a neural network was used to generate a bedroom layout map. ...
Article
Full-text available
In recent years, deep learning methods have been used with increasing frequency to solve architectural design problems. This paper aims to study the spatial functional layout of deep learning-assisted generation subway stations. Using the PointNet++ model, the subway station point cloud data are trained and then collected and processed by the author. After training and verification, the following conclusions are obtained: (1) the feasibility of spatial deep learning for construction based on PointNet++ in the form of point cloud data is verified; (2) the effectiveness of PointNet++ for the semantic segmentation and prediction of metro station point cloud information is verified; and (3) the results show that the overall 9:1 training prediction data have 60% + MIOU and 75% + accuracy for 9:1 training prediction data in the space of 20 × 20 × 20 and a block_size of 10.0. This paper combines the deep learning of 3D point cloud data with architectural design, breaking through the original status quo of two-dimensional images as research objects. From the dataset level, the limitation that research objects such as 2D images cannot accurately describe 3D space is avoided, and more intuitive and diverse design aids are provided for architects.
... Cho et al. used the oor plan of the building as the input data and the semantic segmentation results of labeled various elements as the output to obtain a GAN that can identify and label the elements in the building oor plan 9 . further studies, such as Zheng et al. used vectorized coordinate values instead of pixel maps for deep learning of oor plan layout, and the design elements in the house plan were vectorized into coordinate values to study the bedroom layout problem in the house plan, based on which a neural network is used to generate the bedroom layout map 10 . The vectorized data can describe the nature of CAD data more clearly and have stronger compatibility compared to pixelated images. ...
Preprint
Full-text available
This paper aims to study the generation of spatial function layout of subway stations assisted by deep learning, and train the point cloud data of subway stations based on the Pointnet + + model in deep learning. The point cloud data of subway stations comes from the subway station data of large and medium-sized cities collected and processed by the author. After training and verification, the following conclusions are drawn: (1) It has been verified that the spatial deep learning of buildings in the form of point clouds is highly feasible, and the point cloud format of the architectural space model is fully compatible under the Pointnet + + model. (2) Verified the effectiveness of Pointnet + + for semantic segmentation and prediction of subway station cloud information. The results show that the predicted data has 60%+MIOU (MeanIoU average intersection and union ratio) and 75%+Acc (Accuracy). This paper uses an interdisciplinary research method to combine deep learning of 3D point cloud data with architectural design, breaking through the current situation of using 2D images as research objects, and avoiding the application of "deep learning" to 2D images. Objects cannot accurately describe the limitations of 3D space, providing architects with more intuitive and diverse design assistance.
Article
Full-text available
In recent years, as machine learning has been widely studied in the field of architecture, scholars have demonstrated that computers can be used to learn the graphical features of building façade generation. However, existing deep learning in façade generation has yet to generate only a single façade, without comprehensive generation of five façades including the roof. Moreover, most of the existing literature has utilized the Pix2Pix algorithm for façade generation experiments, failing to attempt to replace the original generator in Pix2Pix with a different generator for experiments. This study addresses the above issues by collecting and filtering entries from the international Solar Decathlon (SD competition) to obtain a data set. Subsequently, a low-rise residential building façade generation model based on the Pix2Pix neural network was constructed for training and testing. At the same time, the original U-net generator in Pix2Pix was replaced with three different generators, U-net++, HRNet and AttU-net, for training and test results were obtained. The results were evaluated from both subjective and objective aspects and it was found that the AttU-net generative network showed the best comprehensive generation performance for such façades. HRNet is acceptable if there is a need for fast training and generation
Conference Paper
Full-text available
In this study, we developed a method for generating omnidirectional depth images from corresponding omnidirectional RGB images of streetscapes by learning each pair of omnidirectional RGB and depth images created by computer graphics using pix2pix. Then, the models trained with different series of images shot under different site and weather conditions were applied to Google street view images to generate depth images. The validity of the generated depth images was then evaluated visually. In addition, we conducted experiments to evaluate Google street view images using multiple participants. We constructed a model that estimates the evaluation value of these images with and without the depth images using the learning-to-rank method with deep convolutional neural network. The results demonstrate the extent to which the generalization performance of the streetscape evaluation model changes depending on the presence or absence of depth images.
Conference Paper
Full-text available
With the development of information technology, the ideas of programming and mass calculating were introduced into the design field, resulting in the upcoming of computer-aided design. With the idea of designing by data, we began to manipulate data directly, and interpret data through design works. Machine Learning, as a decision making tool, has been widely used in many fields. It can be used to analyze large amount of data, and predict the future changes. Generative Adversarial Network (GAN) is a model frame in machine learning. It’s specially designed to learn and generate output data with similar or identical characteristics. Pix2pixHD is a modified version of GAN that learns image data in pairs and generate new images based on the input. The author applied pix2pixHD in recognizing and generating architectural drawings, marking rooms with different colors and then generating apartment plans through two convolutional neural networks. Next, in order to understand how these networks work, the author analyzed the frame of them, and provide an explanation of the three working principles of the networks, convolution layer, residual network layer and deconvolution layer. Lastly, in order to visualize the networks in architectural drawings, the author derived data from different layer and different training epochs, and visualized as gray scale images. It was found that the features of the architectural plan drawings have been gradually learned and stored as parameters in the networks. As the networks get deeper and the training epoch increases, the features in the graph become more concise and clearer. This phenomenon may be inspiring in understanding the designing behavior of human.
Article
Full-text available
Thesis (Ph. D.)--Harvard University, 1975. Includes bibliographical references.
Conference Paper
Brickwork is the oldest construction method still in use. Digital technologies, inturn, enabled new methods of representation and automation for bricklaying.While automation explored different approaches, representation was limited todeclarative methods, as parametric filling algorithms. Alternatively, this workproposes a framework for automated brickwork using a machine learning modelbased on image-to-image translation (Conditional Generative AdversarialNetworks). The framework consists of creating a dataset, training a model foreach bond, and converting the output images into vectorial data for roboticassembly. Criteria such as: reaching wall boundary accuracy, avoidance ofunsupported bricks, and brick's position accuracy were individually evaluated foreach bond. The results demonstrate that the proposed framework fulfils boundaryfilling and respects overall bonding structural rules. Size accuracy demonstratedinferior performance for the scale tested. The association of this method with`self-calibrating' robots could overcome this problem and be easily implementedfor on-site.
Fresh Eyes -A framework for the application of machine learning to generative architectural design, and a report of activities at Smartgeometry
  • K Steinfeld
  • K Park
  • A Menges
  • S Walker
Steinfeld, K., Park, K., Menges, A. and Walker, S.: 2019, Fresh Eyes -A framework for the application of machine learning to generative architectural design, and a report of activities at Smartgeometry 2018., Proceedings of CAAD Futures 2019, Daejeon, Korea.
Necessary Tension: A Dual-Evaluation Generative Design Method for Tension Net Structures
  • M Turlock
  • K Steinfeld
Turlock, M. and Steinfeld, K.: 2019, Necessary Tension: A Dual-Evaluation Generative Design Method for Tension Net Structures., Proceedings of the Design Modelling Symposium 2019, Berlin, Germany.