Scribd
Upload
48 views9 pages

Deep Learning Assignment 1

DL 1

Uploaded by

podalavigneshc3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
48 views9 pages

Deep Learning Assignment 1

DL 1

Uploaded by

podalavigneshc3
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF or read online on Scribd
Nvaxshitha AIGA IARIIE Answers Questions shote AS ES pene deep learning 2 what ove the types of Beep eoncnapiongp? wrat *8 a based leasning? what 18 wneart by hidden unit? Depine Stochastic fog descent? long AMSweY Ques tiong jlang COREL CURES | Explain albout Deep Learning 2 phat ave the Higtosical trends in Heep learent ng 4 )| Explain albout fpocient based leasetng? Jj wovike about Back - propagation Networks 3 | ons A yite about Aychitecture Design? 21g ath3i92 Shost Angwezs Dn oeep | leargoing? peep leasning 18 o Bubsteld of ymodhine learning that deals ust tt alqouitogs tapised by the stvuctuve Of the byain *Deep Leonia Subset of machine leaning puri tg a post Ob Avtipicial trtelligence Tues of Bssp lecseing * comolutional neuval netuosk CONN) Recusvent neusal network RAN) De noising auto encode (DAC) Deep belie® netwowks CDBNS) long Shot — Term memosy (18TM) Gsadient bated learsring > SARRAAE ARAR ERRAS Gradient based learwning 38 0 type of machine learning wm which the optimi cotton algosithen uses Yrasients to update the model parameters eel SOTO A Hidden unit Defers to the components | eet the layers of aes betwee | wput and output units im a Comec Honig System BIGAIA3IIB gxochostic Quadient descent F | amt Seon atochas Be Room deacent SGD) *8 vasiant | ed 3B of the Gradient descent algositnrs twat 13 U3 ype optinont cing machine leasing model’: Gr addvesses tne computational seppictency og | - | cyaditional Bireattow descent metnods jrivial wetght ea h.—.UlU 9 . Long AUER learning = OES eae 3 ji tte applieati- Before we Get unto deep learning » pp 4 ste ons and platfosms , the +i¥st ting th ‘ «UL tntvoduerion to deep leasving tutosial es’ help you understand 48 what exactly is deep leawving: peep learning 18 o Bupteld of wnachine leawn- img trot deals with algersitens inspired by Me Btsuctuse and poston OF the bsain- Deep learning tg a Subset of machine leaning, which 78 a past of awtigicral Intelligence cee) Rctifictal sntelligened Ave netal tneetti eT dicta Iteltiqence 3 the ability of a tate mock ne learning macvine Tmt | to imi intelligent human behas ats System to leas Ten prov RENE tsorn experience automation Learnin ma ini al Application op giqalaanas = ed pee: BIG RAIIB prat UBeS Cornplex algoettinms Ond deep ywal NOES to treaty a model. peebre Cate eee ce Reo an | Q ose evolved branch of peep learening mochine learning On8 uses layess of alppritone | xo psecess data, and trritate the thinning process ,o8 to develop abstractions. iq “ie Wistory of deep learning Can be eocod. | bach to 1943, when walters pitts ond wanyen | Culloch created a computer based model on the neusal we tioosns ef the hurran brain: “Mey Used oa combt nation °} algosttnms and Sg Ss they Called “trveshola logic" to mic the Hoought process. wi | The 19608 | Cece Henny Ske tley 18 given cwedit tow developt 9 the basies of a continuous Back P®opagation wnodel tn 1960 y Busing the la4o's the first AT winter Kicked fn, Khe ‘sesult of promiges that couldn't be Kept: “Me trpoct of this lack of gunding Utnited beta pL and AT yeseaseh, 4 19808 and Jos gigaasi98 a f 4 er 1439, Yann Lecun provid ed the +) gtsation at Bell Labs + neural net os) ; a “hand wsttte niga red’ co seco tals e He demon oe practical combined Convolutional back propagation arte sea Wis Systern OAs eventually vuvnbexs 0} Wand usttten cheeks: Q00 ~2O10 Be AAS Around the Year 000 , The voi hing qrodien® problem appeased +94 was discovered “yeaturemg gormed tr lowes loyess were Not being go> | Se Remeas ihe seciec net cuas * aes descent tg a Common optimization algosttnnn used ty gradient based leaning to update the model powametess. * Merve ove several Vorviants of gradient descent gudhag batch, 8tochastie jan Fete ; ,ard, minis batch’ gra Kent descent Nantishing and % AD ES Splosiog gredients + 7K Nonishing and explody Met WOwKS, * Te e Ber ies dumme. ube, ght tritializ aus, _- NGIA3193 in qunctions these ssaued oetght : ? [

You might also like