Perceptron

Question 1

Perceptron Algorithm

Answer 1

Learns a linear boundary to separate data into two classes.

Question 2

Dot product between two vectors

Answer 2

The dot product between two vectors is the projection of vector a onto vector b. How much of a is pointing in the same direction as b

Question 3

Dot product is positive when

Answer 3

Vectors are pointing in same direction (acute)

Question 4

Dot product is 0 when

Answer 4

Vectors are perpendicular

Question 5

Dot product is negative when

Answer 5

Vectors point in opposite direction (obtuse angle)

Question 6

Perceptron characterized as

Answer 6

Feed forward neural network that can be used to solve linearly separable problems

Question 7

Perceptron Strong Guarantee

Answer 7

If the data is linearly separable, a perceptron will find the hyperplane that separates it

Question 8

3 parts of perceptron algorithm

Answer 8

Input/feature vector
Weights
Bias

Question 9

Goal of perceptron

Answer 9

Learn these weights in a way that correctly classifies the input data into one of the two categories

Question 10

How does the Perceptron make a prediction?

Answer 10

Using the sign function, which determines whether the weighted sum of inputs (plus the bias) is positive or negative

Question 11

Why cant perceptron learn XOR function?

Answer 11

Cant distinguish nonlinear lines

Question 12

What kind of neural network is Perceptron?

Answer 12

Perceptron is a “one cell” neural network

Question 13

As long as there is a hyperplane that will linearly separate two classes

Answer 13

A perceptron will find it

Question 14

Activation function

Answer 14

Take your input and squish them to a small range of values.

Question 15

Output of activation funciton

Answer 15

is input to other neurons

Question 16

Perceptron is rudimentary “”

Answer 16

Reinforcement algorithm, it is rewarded and punished

Question 17

Perceptron Assumption

Answer 17

Binary classification problem
Data is linearly separable

Question 18

Error threshold

Answer 18

If my error is less than this threshold, then the perceptron has converged (done its job)

Question 19

Learning Rate

Answer 19

How many jumps it does, high means the hyperplane will jump around a bit

Question 20

Epoch

Answer 20

One complete pass through the entire training dataset during the training process

Question 21

For a perceptron, each point is

Answer 21

a geometric point in a 3d plane

Question 22

Sign function

Answer 22

checks whether the result of this weighted sum is positive, negative, or zero

Question 23

bias always has value of, why

Answer 23

1, so it is always equal to the weight

Question 24

w defines and is perpendicular to what

Answer 24

the hyperplane

Question 25

Training is all about

Answer 25

adjusting (learning) the weight parameter, until the response predicted by the perceptron becomes consistent with the true response