Study with the several resources on Docsity
Earn points by helping other students or get them with a premium plan
Prepare for your exams
Study with the several resources on Docsity
Earn points to download
Earn points by helping other students or get them with a premium plan
An overview of decision trees, adaline models, and back propagation in neural networks. It includes explanations of key concepts such as leaf nodes, parent nodes, and the architecture of adaline. The document also covers the training process, weight adjustments, and the realization of xnor functions using the adaline model. Additionally, it delves into back propagation, detailing the calculations for updating weights and biases in a neural network with given inputs. Examples and steps for solving back propagation problems, along with explanations of entropy and information gain in decision trees. It is useful for understanding the fundamentals of neural networks and decision trees, providing a mix of theoretical explanations and practical examples. Suitable for students and professionals seeking to grasp the basics of these machine learning techniques.
Typology: Schemes and Mind Maps
1 / 65
This page cannot be seen from the preview
Don't miss anything!
PS
Unit- 2 Decision Tree Learning Date
(a) Page 27
1 Decision Tace- It is used to create a
learning model that can
be used to predict the class ar value of target
vacuable.
training
data to predict the class of a new
example. In decision tree four
of class variable of a new
example (data) we start from the
hoot node of the force. we
compare the value of a goot attributed with the
already store. record attributes on the
basis of comparisons. We follow the
branch corresponding to that value and
jump to the next node.
There are two types of decision tace based
on the type of target functions al- 1
1- Categorical Variable Decision tace. 2-
Outlook A
Overcast
Humidit y Yes
Migh Normal
No
Node
Terminal Terminal!
Node
Child Node Child Node t Terminal Lead Node Date Page 29 A General decision tree structure-
class level, Decision Tree
task in ML.
is represented by a decision tace. *Root Node- It reperevent the entire population sample which get further divided into two or more sets. * Splitting- Its a process of dividing a node into two or more sub node to increase tree.
further subnodes then it Date^ is^ called^ a^ decision^ node. Page 30
which do not split
called branch. our Subtree_._
*Parent or child- Achode divided into sub node is called
node are called child node.
shol 3 Oct 2023 Date Page 31
information contained a random variable
It is denoted by
'F' or 'H' оя
meal wre o f
XI (ht
Chi फर
the
The
processing model of human brain
(1015 links). E * Every
neveron receive information from
brain.
pattern. (^) Date Page 33
ANNs are used in Deep Learning
Stock pouce prediction.
Classification problems example Joan approval system
Autonomous ANN.
Basic Teeaminology in ANN-
Inted connections.
Hidden layer .
Models of Artificial Neurons-
ADALINE Model (
Adaptive Linear Newron
Training Algorithm-
each calculations and simplicity weights and bias must be set equal to o and the learning rate must be get equal to 1.
are adjustable Step- 2 Continue step-3 to 8 when the stopping condition is not towe.
Step-3 Continue step 4 to 6 for every bipolag
Step- Activate each input unit as follows (^1) X
(X ) Why OP Activation Function ry
Compariso n with olp Ni=Si i= 1ton.
Step- 5 Obtain the new input with the following relation- you
Yuet = ≤ ni wit b E (^) Date Page
w to obtain the final output.
1 if Yin ZO -1 if yin Co of ADALINE 36 50ct (^2023) Date
Realization of Page 37 XOR junction with the help model.
XI X t O O O 1 X, X2 = x1· X2+ X1. X2_._ X1 + x2 = X1 X2 +x1x2. 1 1 1 O replace with -1. Iteration -
2 22 + Yin 7 Winew W2new b new error
1 1
2=- + 22 =- Yin = b + {xili. Yin = 10+ N, W1 + X2 W Yin = -1x0 + (-1)x0.. Yin = The Winew = wiold + Awi. winew = wiold + a (et - yin) ni W1new = 0 + 1 (-1-0)x- 1.
bne w = 1 bold + of (ut-yin). = 0 + 1 (-1-0 ) =-
-> E (^) Date Page 38 Date Page 39 2 =-1, 22 = 1. yin = b+ ≤ni wi Yin = -1+ (-1) x 1 + 1x1. Yin = -1 -1 + 1... yin = -1. Winew = 1 + 1 ( 1 - (- 1 )) X- 1 = 1 ( 2 ) =-1.
W2new = 1 +1( 1 - (- 1 )) x 1
=1+ 2 3. bnew = bold to Cit-Yin) = ÷ 1 + 1 (1-(-1)) =-1+ 2 = 1
21 = 1 , 22 = - Yin = 6 + Eniwi. = yin = 1+ 1x(-1) + (-1)x3. Yin = 1-1 -3. Yi = -3.
W1new = -1+1 (1-(- 3 ))x 1. =-14. 3.
= 3+ (13) X- 1 = 3- 5- bnew = =
N1 = 1 , Jin Yin = bold + a(t-yin) 1 + 1 ( 1 - (- 3 )). =1+ (1+3) =1+ 4
22 = 1. = 5 + 1x3 + 1x- 1.
= 5+3-1. Yin = 7. Winew (^) = 3 +1-1-7 ) x1. = 3 + (-8).
=-5+ (-12x-1). =-5+ 12 = 7. W2new = -9 + 1x (-1-11)x- 1. =-9+ (-12 X-1). =-9+ = 3 bnew = −3+ 1x (-1-11). =
Mean errou = 968. Iteration-
~ 1 = −1, N2=-1, Wlord =-29, w2 old = - bold = -11 (^) old
Winew W2new bnew e x= 1.
21 22
yin y -1 -
1 -145-
1 135 29 13456
1 -175 - 107 - 1 1 -1 271 1 -165 -313- 73 984 205 30976
mean evro≈ 31048- Iteration- 4 UTJAGA W1old = - Woold = -313. bold =-67 α= 1
х+ X
1 1 -627- -381 727 149 1 -1 1 1 1
11 120
Repla ce O with -1. a = 1, bold = 0, Wald = 0, W2014 =0.
Iteration- 1 mean eager = 968.0 (^) Date
Page 43
α=1, bold = 11 11 , Word= 29 , W2 old= 51 Iteration-
N N
-19 87 енном 5776
N 22 -
4- 2+ 126 2
-1 1 -411- y Winew W2new bnew -247 -99 енном 479 169744 1 1
