Home : Book details : Book description

Description of  DEEP LEARNING WITH THEANO

D lrning i a ubt f the larger fild f mhin lrning that attempts t mdl high lvl btrtin in data in order t vtl improve rfrmn in bth supervised nd unsupervised learning. It hiv this b uing multil lr f "rr", each f whih contains a t f non-linear transformation functions that lrn rrnttin within th dt. Suh n rh i motivated directly by hw the brain i thought t work. Ovr th ur f ur liftim we lrn but iml ideas nd thn u these iml id t frm hierarchies f mr complex id. Deep lrning i about applying thi rh t mhin learning tk. prerequisites in order t get trtd. In particular n huld b familiar with th bi elements f linr algebra, vtr lulu (gradients, partial drivtiv) and probability (mximum liklihd estimation). In dditin t th mathematical rruiit it ruir a good undrtnding f object-oriented rgrmming and, for ffiin ur, a bi familiarity with the rtin of a GPU. Hwvr we will tr nd intrdu mn f these nt they r ndd. In these Bk we r ging to find ut wht deep lrning i, why it h recently bm ulr, hw it wrk and hw we n l it t r f untittiv finance in order t imrv ur model results and portfolio rfitbilit. W will mk u f a Python librr called Thn nd GPU, whih will rvid ignifint increases in trining rfrmn.