0% found this document useful (0 votes)
13 viewsAiml Module 1
Artificial Intelligence and Machine Learning
Uploaded by
Lakshmi LakshmiCopyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
0% found this document useful (0 votes)
13 viewsAiml Module 1
Artificial Intelligence and Machine Learning
Uploaded by
Lakshmi LakshmiCopyright
© © All Rights Reserved
Available Formats
Download as PDF or read online on Scribd
/ 15
Definition: Artificial Intelligence is the study of how to make computers do things, which, at the
moment, people do better.
According to the father of Artificial Intelligence, John McCarthy, it is “The science and
engineering of making intelligent machines, especially intelligent computer programs”.
© scanned with OKEN Scanner Branches of Al:
{isto ances of A pve low Howenr sme branches ying eats a
‘nc Wt them ye Some os maybe ep os ones pe i an
‘ance
Logical A1— tn gnc he cs fhe pie taste which ita at ans palate
tl wesoued by srses of waatel lied ngs: The pram deco
Search — Arif nlgece prin fin ceamine ge suber of poetics ~ for
cle menace ne nce a we Ping ra, Dh
Patten Recognition — When appr aks beatin so kin i fe pln
‘occur wat secs wt pate Fee xan, avs ph mat UO ah pate
‘yer and a ut eome ore Gal foe, Mev comes pore av ie ata
{tnpnge te,» chs poston ohh Matyas event The mie cos patent
eu ent mts th he pe ptr that an bon lh
Representation — Usaly pugs of mattis fg te we test thefts abt
Inference — Ores can te nr fom some fate Mathai Ink satin 6
‘lion fr ae pup, bt ew med auton rece hve Best al 19
{heap sce he 19% Th spt Lind fremont son stank rng
‘nic concasin oe ined y dt he eneishn ean he witha he
Sone det Fer, hn aon th can
anclson may have foe widen th cnsiacs fhe novstonn har the
‘cing Neal lpi fenoning toon ato encase be ve
frac of promises te monte hceasing Teton eh pomiesCensespon
smth menenoine eanoie.
Camon ene koe and Ressoning — This is these ia which As fares am the
una even pia he at tha has boc a ae eset tea sce he 19808 We
thee hn bee cemsierbl peepee in dvckpig snl mtn rewonig
Leorning frm experens — Thre re ane res eesti i lean Progra
"inom ium wha eso” beuo hi foram ean eyes runt ig
Syste ae los ll tse ery led beso eps na
Panning — ning str wih evel fc abot he worl (spilt ab he st
‘taconite aa th ania stuacon an scent of goa From hse, peng
roztams gore» sat) frst el Inthe sean nese Sate
Epitemalogy — This ica wady of he kinds of knoe that ae rei fo sing
bts ia the word
Ontology — Onno the study ofthe Kind of things ta xin AL the potas and
Senne: del with vou Kinds objss anne sty wha hse Kind re ad Wt ie
as proper a ally astumed importance ro he 19908
© scanned with OKEN Scanner Task Domains of Al
Mundane Tasks: Formal Tasks
Perception Games : chess, checkers etc
Vision Mathematics: Geometry, logic, Proving
Speech properties of programs
Natural Languages Expert Tasks:
Understanding i Engineering (-Design, Fault finding,
Generation |
Manufacturing planning)
Translation Scientific Analysis
Common sense reasoning
Robot Control
Medical Diagnosis
Financial Analysis
© scanned with OKEN Scanner Applications of Al
AL has applications in all fields of human study, such as finance and economics, environmental
engineering, chemistry, computer science, and so on. Some of the applications of Al are listed
below:
© Perception
= Machine vision
@ Speech understanding
= Touch ( tactile or haptic) sensation
* Robotics
© Natural Language Processing
a Natural Language Understanding
Speech Understanding
m Language Generation
= Machine Translation
Planning
Expert Systems
Machine Learning
‘Theorem Proving
Symbolic Mathematics
Game Playing
© scanned with OKEN Scanner 234 State Space Search
‘Nae space eres ble inems.saes and pers th cha sts,
‘Rete ue const
1 A revesetion ofthe snes th system canbe in. Fr example
oar gamete oud yes cue ue of he ah
‘+ Asetsfoperarars ht an chang este aote sto bord
‘Bane theyre ae th el eves oman vn sate OMe he
{pen repeals Poza can aterep Se
reyes he nw sate
Aninol sae
1 Asctr/nat stares some o hse maybe dstable, tes dsb
‘Tisset len epeseedinpicly by a rogram
2.2 The Water dug Peabo
Ins pobien, we
e1wo ascii four thee our de a manu of feu gallos
amass of he plo war How cn att galls eaten
‘These space ia st of stnge sits ning the umber gallons warn th pio
Lis tants eeu, thee whee foe =A 1,2, Toedand hee 0.30}
‘he start stati (0,0) sh ale 2.) whee m yb ny ba imi thee
alg eam 049 3 ss wae oem. Tec and ta shuns hs mes mane
ue haste wan jogs sing tao pba Thea
rection sf sling ts problem as shown bw
teenie,
tree oer <4 (Steer fom ap
2 ten ets} {Gur 3) wee oma
4 tine eo ow >0 {Crdece copy our ne dain
tia ce > 0 {tar vey es ne ain
5a te i+ ect (tur tree Dems eo
6 tn te i + esd (Gor +t) mayen
ec
200,010 tee >0 {tres ey cinta ou
tie itr {0\an ey ort tee
Sai €2.d)empy tee far
Go {0.2.empy tain
Tae ce itor 4 {tec rit ur in
12 (ie, fou irtee <3
‘vee es ur ana seat
henbekow hurts ne
(i 22 Proucon Bef the
Gaon in Four Jue
(e238 One Slaton wth Water Proton)
© scanned with OKEN Scanner 2.4 PRODU
ron
‘TEMS
Production systems provide appropriate structures for performing and descr
processes. A production system has four basic components as enumerated below.
ng search
‘* A set of rules each consisting of a left side that determines the applicability of the
rule and a right side that describes the operation to be performed if the rule is
applied,
‘© A database of current facts established during the process of inferen
+ A control strategy that specifies the order in which the rules will be compared
facts in the database and also specifies how o resolve conflicts in selection
I rules or seleetion of more facts
module,
‘The production rules operate on the knowledge database, Each rule has a precondition—that is,
cither satisfied or not by the knowledge database, If the precondition is satisfied, the rule can be
applied. Application of the rule changes the knowledge database, ‘The control system chooses
hich applicable rule should be applied and ceases computation when a termination condition on
the knowledge database is saistied.
© scanned with OKEN Scanner Example-1: Tic-Tac-Toe
fst approach (simple)
The Ti-TaeToe game corsss of a nine cloment vector called BOARD; it represents the
umber 1109 in hes rows
ay 2
la}sTe
T/S]9
‘An elma contains the val fr blank, for Xand 2 for. A MOVEABLE vectr consist
6 14488 element ("and is neds what each clemea isa nine element yet. The costes
fofthe yea especialy chseh to elp the aati,
Te algeithm makes mvs y pursuing the flowing:
View the yet aya ley number, Convert to decimal amber
2 Use th decal number seamen MOVETABLE and acer he vector
A. ScrHOARD tothe veto nding os the Hoard Inks er the move. Tic pci
‘pst inte Bul ih sever daaragey Te aks more space and feu sna
fit aot the decimal numbers. This method i psf to this ame ard camot be
mle.
1.2. The second approach
The stuctue ofthe dai as foe bat We ws 2 fra Bank, 3 fa a X and $ for a8.
A sarah called TURN iis I Yor the Rist ve and 9 fr the ast The alg esis
thee stn
MARE? which tur $i he cere agai blak; here it tums any blank on
‘eornr square, ie 24 6 & POSSWN fp) Flan OA player p cannot win on the Bex! mee
anton fetus the ner of tho squat ia gies a wn ae
{ar 0, an the wining move fh older of the blank, GO im) makes # move wosqaae m seing
BOARD[a}to Sor.
Tas lg is owe ave and thes loge bis mone ecient in storage which
‘compenses or its longer tine. depend on the programme's sil
La The
approach
“The sete ofthe ta consists of BOARD which consis nine sent sect, ist oF
toa postions ht cul et forthe next move aaa nh representing an enalion of hot
‘he bd potion ess tan lina win fe the payer to move
sarin lake shel ta mates dace on the ot move by deciding which hemos
‘romising move othe mes tle move any stage would be and sls the ae,
‘Consider al posible mows ar vepis hat he program can make. Comins process foe
slong ie permits et winsr emerges, ben chose the move leads Ihe compar
ogra wining. psi the shrest ne
‘Acaly hsis mos dic oprogra by good int at sas at the technique can
te evend ain any game, Ths method males rely fewer leas on th presamse fe
‘ofthe ge tecigu but the vel pms state must eho the aise.
© scanned with OKEN Scanner Mill Climbing
ill Climbing is hewiste search used for mathemat
Autti Iligence
Given a large st of inputs and a good heuristic function, tres wo finda sufficirly good
solution to the problem, This solution may not bs the global eptimal mana,
‘Inthe above delnition, mathematical opinzation problems implies that hill limbing.
sles the problems where we need! to maximize ot minimize a given ral function by
‘haosing values Fm the given inputs, Fxample-Lvelling salesman gvoblem where we
faced to minimize the distance traveled by salesman
+ ‘Hust search means that tis search algorithm may'not find the optimal solution to
the problem. However, wil give a good solution in reasonable time
‘+ Avheuristie fuetion is function tht will rank all the possible alternatives at any
tramching step in search algorithm hase onthe avaiahle infomation. It helye the
gorithm to select the het route aut of possible routes
Features of Hill Climbing
1. Variant of generate and test algorithm It is a variant of generate and test algorithm. The
‘generate and test algorithm sas follows
pimization problems inthe fel of
1. Generate a possible soltions
2 Test 10 se if hist the expected solution.
5. fhe solution has been fn quit else go t step 1
Hence we eal Hill elimbing 35a variant of generate and test algorithm as it takes the feedback
{rom test procedre. Then this Feedback is wilized by the generator in deciding the next move in
search space
2. Uses the Greedy approach At any pont in state space, the search moves in that diecton
‘only which optimizes the cost offanetion with the hope of ining the optimal solution st
the end.
1. Simple Hill ctimbing: It examines the neighboring modes one by one and selects the frst
neighboring node which optimizes the curren cost as next node,
Algoxith foe Simple Hill lining
Step 1: Evaluate the nunal state. sa goal state then stop and retun succes. Others,
‘make intial state as current sate.
Step 2: Loop until she solution statis found or there are no new operators present which can be
‘applied ro curren tae
4) Selecta sate that has not been yet applied to dhe enrrent state and apply tw produce a mew
1) Perform shese 10 evaluate new state
{-dfthe curren state 1 a goa! state then stop and return succes.
1 ft is beter than the current state, then make it current state and proceed further
fi Ifa i nt better than te current state, then eomunue inthe lop tna soation 1 fo
Shep 3: st
© scanned with OKEN Scanner 2. SteepestsAscent lll ctimbing: I fst examines ll the neighboring nodes and then
seletls the node closest the selon state as nest nod,
Sep 1
Sip 2 Repeat these steps uni solution found or curren state des not change
{Let arg” bea sate such that any suecesvorof the curren state willbe beter than it
{i Jor each operator that applies ta the curren stale
‘appl the nes operator an ereare a mew state
evaluate the new state
6. this state is goal sate then quit else compare with “target”
if this state is heer than target’ set this sate as target
€ sftarge is beter than curren state set currem state to Target
Step 3: Exit
3, Stochastic ill climbing : I does nol examine all the neighboring nodes before deciding
‘hich node a select. just selects neighboring node a random, an decides thased on
the mount of proven in ht me wheter Wo move tt ig oo
examine another
State Space diagram for Hill Climbing
State space diagram i a graphical representation ofthe set of states our search algo
each vs the value of or ebjctive tutions ftion which We wish to mani
Evaluate the initial state {fi i goal state then exit ele make the current state as initial
objective funtion
obama
loca masini
sate space
Different regions in the State Spee Diagram
[Local masimam: Itisiastate whichis etter than its neighboring state however the
exists state which i eter than (global maximum), This state is beter because here
‘value of objective fanetion s higher than ts neighbors,
“
© scanned with OKEN Scanner 2. Global maximum : 11 is the best possible state in the state space diagram. This because at
this state, objective function has highest value
3. Platewa/fat local maximum : [is a Mat region of state space where neighboring states,
have the same value,
4, Ridge : Itis region whieh is higher than its neighbours but itself
kind of local maximum,
5. Current state : The region of state space diagram where we are currently present during
the search.
6. Shoulder: Iisa plateau that has an uphill edge.
Problems in different regions in Hill climbing
ich the optimal’hest state(global maximum) if it enters any of the
aslope. Itis a special
following regions
1. Local maximum - Ata local n
worse than than the current state. Since hill climbing uses greedy approach, it will not
imum a
neighboring states have a values which is
the worse state and terminate itself, The process will end ev
solution may exist.
‘To overcome local maximum problem : Utilize backtracking technique. Maintain a list of
visited states, Ifthe search reaches an undesirable state, it ean backtrack to the previous
ation and explore a new path,
2. Plateau : On plateau all neighbors have same value , Hence, itis not possible to select the
best direction,
To overcome plateaus : Make a big jump. Randomly select a state far away from current state.
Chances are that we will land ata non-plateau region
3. Ridge : Any point on a ridge can look like peak because movement in all possible
directions is downward, Hence the algorithm stops when it reaches this state.
To overcome Ridge : In this kind of obstacle, use two or more rules before testing. It
implies moving in several directions at once.
n though a better
© scanned with OKEN Scanner 6° algeitam :-
flrs Yen lthtn)
Stepr Stort cotih pol A
raul Band F
re) z6teelu AOR
Flr) 2346 =@) -. AF
0
Shovited
stp, ADF-OH
ADF OG
Flu): 347432: 19 .ADFOH
pla)e 3414S 2 @ + ADF OG
u
Sepu- Qopoaot
How thew & no det Eon
© scanned with OKEN Scanner Ple)2APFOUDTOE = HIF SASH IANS
EU: APF PHOTOS = THIET 240-0) > Shortet
pld)s APFP 4D TOM 2 Bt lt FH 2UAIBZAI2
Steppe: F(T): AVFIGITOT
wBinal stele, thee & no deton nbole
©® Scanned with OKEN Scanner jy (nets
Ao sltbe pe
6
yao 4 ice) - Whe S
aCe) Hie) ba 0ST KCDY= 12 HB=
Kinde
¢ (Ete) = 462 F Cr istest
Yeaets- age thr )= Weed sroplace by #
saated 4 propagate bac
Catevtake | CA- B) = gCosy4 hpy= bee’
then neplace 5 "by
sok phd TAT, Fis smalbest one
(e-@= aed
\ 2 ‘ a.
Slee a de ee Sak Aas ye alneody
J Co-s)= ro
Propegcde 3 vegies 2 vs)
Propegede k gek
\Er-e-p)= ge Hc crepes
= year eS
© scanned with OKEN Scanner Heuristic search
oper time rather than a complete solution in unlimited time we use
istics. "A heuristic function isa function that maps from problem state descriptions to
measures of desirability, usually represented as numbers’, Heuristic search methods use
knowledge about the problem domain and choose promising operators first. These heuristic
search methods use heuristic funetions to evaluate the next state towatds the goal state, For
finding a solution, by using the heuristic technique, one should carry out the following steps
1. Add domain—specitic information to select what is the best path to continue searching along
2. Deine a heuristi function h(n) that estimates the ‘goodness’ of a node 1
Specifically, h(n) = estimated costo distance) of minimal cost path from nto a goal state
3. The tem, heuristic means ‘serving to aid discovery’ and isan estimate, based on domain
specific information that is computable from the current state deseription of how close we are to
goal
Finding a route from one city o another ety isan example ofa search problem in which
different search orders and the use of heuristic knowledge are easily understood
1. State: The current city in which the traveller is located.
2 Operators: Roads linking the curren city 10 other cites.
3, Cost Metric: The cost of taking a given road between cities.
4. Heuristic information: The search could be guided by the direction ofthe goal city from the
current city, or we could use airline distance as an estimate ofthe distance to the gos
Heuristic search techniques
or complex problems, the traditional algorithms, presented above, are unable to find the
solution within some practical time and space limits, Consequently, many special techniques are
developed, using heuristic functions.
+ Blind search is not always possible, because it requires too much time or Space (memory).
Heuristics are rales of thum; they do not guarantee a solution toa problem,
+ Heuristic Search i @ Weak technique but can be effective if applied correctly; i equites
domain specific information,
Characteristics of heuristic search
+ Heuristics are knowledge about domain, which help search and reasoning in its domain
+ Heuristic search incorporates domain knowledge to improve efficiency ver blind search
+ Heuristic isa function that, when applied to a state, etums value as estimated merit of state,
with respect to goal
Heuristics might (Kor reasons) underestimate or overestimate the merit ofa state with
respect to goal
¥ Heuristics that underestimate are desirable and called admissible
+ Heuristic evaluation function estimates likelihood of given state leading to goal state
function estimates cost from current state to goal, presuming function is
(© Scanned with OKEN Scanner Bente Ce
1 Seach stag. wish the highest ayer dsison wees searched compte tore
preceding to tenet ae cll Bra corch (BFS)
TTeth tegy, no viable solution a cid and therfore guranteed that an optimal
{i Othenrse ald the ew sie egnorLST
Advantages
1g storage eqaeme
Deptt sear
{serch tate) tat extends the caren path os faras possible before haciackng wth st
Tithe sting tae a goal state, gt and str sce
2 Onherwne, do the fon nl see ae aed
1 Generate usnceester Eto he uting ste 1 ler are more succes, then sgl il
1. Call Def Search with E asthe startin sate
Hsecesretured signal sucess thers continue in the
Ite storage requirement diner with ee pth
ass programmed ncn al sack ds ros the work of manag sts the
[Pay ind sub-optimal slton feta is deoero more ay than the best stor)
2 Incomplete. without a deh bound, may ms Tn sion even Foe es
2.42.3 Bounded dept sear
‘spt search can sper! much ime (perhaps infinite tme) exploring very depp tat
