Beginning Anomaly Detection Using

IURP

VNOHDUQPHWULFV

LPSRUW

URFBDXFBVFRUH

GHI

WHVW PRGHOGHYLFH[BWHVW\BWHVW 

SUHGV >@

\BWUXH >@

6HWPRGHOWRHYDOXDWLRQPRGH

PRGHOHYDO 

ZLWK

WRUFKQRBJUDG 

FRUUHFW 

WRWDO 

[BWHVW WRUFK7HQVRU [BWHVW FXGD IORDW

\BWHVW WRUFK7HQVRU \BWHVW FXGD ORQJ

[BWHVW [BWHVWWR GHYLFH

\BWHVW \BWHVWWR GHYLFH

\BWHVW \BWHVWVTXHH]H 

RXWSXWV PRGHO [BWHVW

BSUHGLFWHG WRUFKPD[ RXWSXWVGDWD

WRWDO \BWHVWVL]H 

FRUUHFW  SUHGLFWHG  \BWHVW VXP LWHP 

GHWDFKHGBSUHG SUHGLFWHGGHWDFK FSX QXPS\

GHWDFKHGBODEHO \BWHVWGHWDFK FSX QXPS\

Figure B-48.  The testing function. Since there are no data loaders, the testing sets 

must be converted into a tensor and passed into a GPU before being able to make 

predictions on them. The AUC score is then generated along with an accuracy 

value

The rest of the testing function code is shown in Figure 

B-49

.

appendix B   intro to pytorch

---

405