—rssignment ~ 2. Detecd on om i Y Qt Determine tu decision rule for the mess ‘ol Estimation theay, Ho ! f (vin) = je? » Fer gz° IHo } °, ow He Lgl) = et! FEI ¥IH) BS ow with (oa a5 using a) Probability of ener critenion 6) N-f Crttemom ua P(e) =O ©) min-max cxtterton: You can assume puleable costs 4) an each case , find tH ote Coverage) probabdlity e ener: QO... Find the clecision ede Using Nop Crittenden efor the measeg Ho + Fame “4 1, O29) on oo, wo 4, Uylh) = e°, Zzo 1H) Ox oe ea uot P(B,]He) = 0-2. Determine a) a valu of fy Auch thet the Frobability of Wer deetsion rule wilt be the Ade as the Heyman ~ feanton decision ‘rate found previowst7g-y ard b) cAscuming tho the messager ant equally get of Costs puch that the Bayes deetsim alec equal +e he N-P rule. Joxely «© ule 4O. Consicter a binary clecision prrblem uth Hef (aIHe) = FCI. -Qewes ° 5 on Dw, °, ? x Ho (gin) “f Z(s-¥) 5 -\ege2 a) Determine Mee minimum probabibity 4 eur decision rule f had Deoteaminn the vf clecisin rule fer fhe of ©) Determine the max _antn decisin vule for Cyop= CH = O and 96,, 2 Corr d) Determing Hee 4oreL probability yor fer He deesion rule va parts a, ®, md W- 3 fer the costs ¢ part (ed. Plot minonurer Boys wisk. as a function 4 R fer Ose). @ Q@y considea a bin decision prvblem ithe vo: FCT He) fe aby , osf eT nt f Cath) = f 2 sity + O89 £7 Gnd the Nef decision rate for PCPIH) = OlGS concider the fellewing bin = hypothe _ ing. bineng hypothesis sestingg He: JeN Hey Ye can where sand N art stetisH cally tnelapenct et wrondom Vartobles thy probability density fuckers f(s) = i 2, -14s<) Op oe Soom Ga tee ow: oy) Set op the LRT ond deter ne He decision wegres when Vee Sau get, 9 nee B) Find fe amd fy fe 1 apres vali 1 pat @» ) Sketch tHe Roc gs ™ conditional densitg feunchions Canes ponahing to the Hypotheses Ho & HW, ant Gove 2 \9) -2 1am f - Fat * ae 2 Fy KK CqiW= +e & a) Find He LRT and dlectsion hors + ; Find tee fe and £0 Ceaedng. Ate eninienurn preety 4 ener and f= 2s o) Disusse He por fermamce, ae min Conk ascents as ie pot G)> yu pased on tee w-f tert jemag test For te 3) Deter tes ale cision dor a Fp = OdQT The random vVanishle x 4 allo, o7) . Mn Peueved irre, dnt Of He Hoe non Linear tram Hol Y= x3 ao Ye ee find LRT: Q.@ the Randorn variable tas ON( 9m, 9) 4 and Ho! Ys *™ Hol ve ta Find LRT- G9. Th obsavatio ia cletine? as Hi Ye Bm+ Het f+. 5 amd G> Varcances, band n ae win zene mean and & weppeetiely- a) Find Hee CRT aed b) Draw He ROC ° ©) fissure that te two hybetnuss ans equally Sivuby. ose Atta Cxiterion 4 minimum prtbebeldly or cour: hat 4 feR. Qe bet an, S » woe YSR* Hy aye TOAD) and ba Ber =D, amd el : rN) Find the LRT, #) Find Hee Roe.Qu The obcewed random Variable, Y pas a Gaussian Penalty on the THrece hiypotherco x) r re Fyiy, C8 = (SS jy te tes Rm keas where the foramen values on tee thee hg beivcscs an, Hoy mse, MAG Ha } mom, B= G mes mee, Beg PAG) wy amd he criterion 42 AE ybotheren ore 4 minimus fredepeliq ef oer: cD) Find Hrs Optimum Bowes tat. 5) Dear the decision eqns Sy cfer On 22 2o% Op 2™ omnd — Campute Tu prbability of ener + - rah, nk let Ge bet Sid | 8, ow 2-N) Uh [eee Fyn, Cat? } a? o; O.9- a) assume Hoot Nt ests at iver bY G2 2Cjiqg 70 Cos= Ci =O 2) Find Ha minimat test of He Versus Hy, amd toe aves pending amtndmay TSI d) Find tae Nop test ct Ho Vem wih feed. Find the Garreeperdurg. fron of the tert:a. Quppose We cbsewe Q vandon vaniale ¥ Given by. Yo N+ OA Whar @ Oo eirw ood, A ow Fived Mumber between 04 2, amd whore N 4 a-vandon variable that hee a oniferm densi onthe drtevel C1, 1) 6 we Wrsh te dleeiclr behween toe hafpotaeses Ho: =o ond Ht G=!, a) find tie wp cletisien wale fer fe ranging from 0 +51. Find the Power of n-p wule ara Feunchiow Paramutr ae Skebeh sti ROG 9 oY he amd algo, donaider Hye Fellewtng- four of Cancevntng- a sequmcee Y, Te ~~ yandem Variables y Ye ow Ne BD RET py 2 Kel” Exppotrere> Yw a Tndapenclent Ho! H,° Ye oN (4 herr jus, &, 5%, Lb, ant Knewon Corstot: CRT can be exprcced as a fumchon of a) Shas thet bo", and ot, aud the quantibes Hue forameter Mo, HM, yoy yn zy A om & Yeb) Describe tee N-P test for Hise tue cases (kee Ly & oe Ly GKrsary and (B= 57 £ My rhe) ' °) Find +h, thretuld and Roc's fer the Care Mor Uy > G2rBr ate NN = Buc Cemsitern He hophottoeres “+4 Buy wth bly rls 0 and OLE = gr 7O + boss then, extet- a ontformaby- more pooufel tert af there hypthece onde, the asccmptten pret fa enn ond mt? tf Le, Find a | how tat Ht a ome. tf nob, Show ong. ana Find the generalized ART- * Q b) othe Qi Consider composite Figpos” speating problem ye Kae soavettoms Ut ty ---NK? hee Ye 2 R=! K art independent Gaussian vandom Variable tte & Kno Vartance or}. The mean mj »j=o,! One = on He ™ mys vO Ho! Mp =O a) Does a OmMF test exist ? b) rf fee oros and MM = 1, osing most poor fuk test, find the Areellest valet sy K shat oll quasentee a pace Jreste han 09:om Consider the binant decision prrblem ote Srjngl? = (ed yr Sat tO * dy voeyes onal for 4> a) calculate minimum Jrrobant dit of eur teat win Feo | tty prvbabllity ef ror Costs: whe NP He ¢). mine mag teat gu8 Find +e tdeel ~ chsewer eleeisio rule y oS 1 = Hos £(y!He) = i é "y Ht FOO: oe e é. tn team Assume Pps te Find He wcesultiny 4 “ @ (> {> os HY dy Also find the — min— max clecision yule+ Assuming {deal —chservea Costs.©: O19 let f(y) = 2 MF FRO mee FOUND v3 420 4) Sim decision rule minimizes fe ond Satisffes fez 1 '9 b) what decisim we Acleets the most frobatle message VF ye 1b omd y= &. ce) if the costs ME Cyp= 0) Cue Cop= 2, Go=S assigned , and if f= feds Thott 4 +he minima Cost and haw can db be achiewd & Q.22 Suppoce Heb oy product of the Aigrat A and a Xero mean Gaussian yanden variable m. tf Spel ond Sie natio text and degentbe Zp ond Zy, sre decision vegtors 21 Supbese Ate observed random variable y 4s Gren by ~ Y= => where S Lamodelled af clo 5) and N aCAl(o, % J > a) fexform He Let Hest. 6) pecembe the decision regres Z and Z,0:22 consiclen thre bin ame problem here pO Y= 2Cl-¥) , 04d) p(¥lte) = 24 » of9 <1 ond Ci Coot Cee ly Cor = 2 expected Bayes rk BasG fem chon ¢ a) Plot the rule Othe orc whlch 6 if whe decision gre minimum 8 Shen f= s. = 23 Plot onthe Bare axis ce) Kepeat (9 fa fie #3 ple a) okereh se core mmhimum © fer each @. o. max test and Plt os fer fovman ton “Hae Kame axls 9.23 fem fique 4, a) Fined + Bayes mule as afumchen oF » by Find amd Sketch minimum fe oe a fumchon Yh 2) Find thee and plot 44 R onthe sare Aer oF nner Fyn V9 fault? enfne ma x decison vule ple ° z ¢ Figua ti feoblem 25.O24 From Figasad , a) Calculate Hee mAINiMUM probability 4 error clectsion Yule ween ed. by Find the ratrimum Ooges Cost ule When = 3» Coo=® Soe 2, Cor =% Cnz2. finns f(rlm) Figues 23 Problem 24- 6.98 Fer tet decision problem wih - f(t dem 4) finn = Qt a) Find the N-P test for fe~ ine EB) Find thy f@ clecision ye if h- 2. txhet> is minimum fe 7 O Ue fe Cocts and dind the wrtn ma x deesion Yule. What Ww Hh atcoclated =o min- may Bertos Cost?Q.26 Qt seems reasonable —Hrat fos “ates est Costs the min~ may clecisim rule eile omesperd ste fee d+ Shaw thet cthic ctatement 0 False in general 5 vce a Counter example Af you cosh. Onder het Condttios on f (yit) and L(1Im) toil) this Statement be true ? Quit Consider He observation Yr SN where Na cn (0, ) N-p test wih fp = 0:20- spot So S amd Si=1, Find the 628 Consider He ed decision problem (440-70) 2 Le YI) = = a e Ee A(y JH= = r a Find the welatonalp beteocen fe = Ol and He wrinimum fy = °6.29 Fos tthe Following, dlecision prvblem formas ely oe f (YIM) = blyrel » ig) 22 if Coa C1=2, Crom % Co 4. ord B- A. Find He Bases decison ule. Using the Aare Bayes costs, find the wine max decision vule- Cind the eptimam Saye Cost in bath Cased: 8:30 Considers Hee binary olecision problem wh Ho! AL IHO = ge? ’ y7° a"? s d#o a=2Z e >» gee HY Hh) = i £Cd)H) } set, Yeo tet f= 4) a) find fe dlecision rete - wy find Nop mele eth Geor <) Find mine mar aleaiston veille, Laing Het conta as per part f- d) qn all tre cater Find te associated probaldle of wer assuming Het fox,Q.3 ; m Jf (41He)= 1, o29¢! f(vh)= e " , Yee Gind tte KOC: Q:32 Considea & binary Preble at. F (vite) a e go? L ite £ F (yim) ater find the dle cision whe rf fo- +: