Neural Networks

Question 1

node

Answer 1

receives input signals
processes them (applies weight and sums them)
passes them through an activation function (threshold)
outputs value to other nodes

Question 2

layer

Answer 2

layer - collection of nodes:

input layer
hidden layer
output layer

Question 3

weight

Answer 3

numerical value that determines how strongly one neuron can influence the subsequent neuron

Question 4

activation function

Answer 4

function has input and then based on the activation function, you decide what the output is (whether it is activated or not)

Question 5

bias

Answer 5

constant, single value applied to the activation function - might or might not change the output

Question 6

supervised learning

Answer 6

type of machine learning where the model data is trained using labelled data
input output pairs where the correct answer (label) is known

Question 7

example of supervised learning

Answer 7

email spam folder
trained to recognise emails as either spam or not spam

Question 8

unsupervised learning

Answer 8

type of machine learning where the model learns patterns or structure from unlabelled data
no predefined outputs or labels
model tries to find hidden structure in the input data

Question 9

example of unsupervised learning

Answer 9

suggested items on amazon
‘people who bought X also bought Y’

Question 10

types of machine learning

Answer 10

-supervised learning
-unsupervised learning
-reinforcement learning

Question 11

testing/ validation

Answer 11

train network with a set of data and test with another set of data to see if the network can predict or behave correctly or not

Question 12

perceptron

Answer 12

simplest artificial neural network
1 input, 1 hidden layer, 1 output

Question 13

what affects the bias value

Answer 13

electrical brain stimulation changes the bias value to excite/ inhibit neurons

Question 14

process of artificial neural networks

Answer 14

-inputs (x1, x2, x3, x4)
-multiply weights by inputs (x1 x w1, x2 x w2, etc)
-sum together
-sum bias together
-function: based on input, which creates the output

Question 15

what does each axis correspond to on a graph of the activation function

Answer 15

x axis = input
y axis = output

Question 16

reinforcement learning

Answer 16

type of machine learning where an agent learns to make decisions by interacting with an environment
the agent takes actions, receives rewards or penalties, and uses this feedback to learn a policy that maximises its cumulative reward over time

Question 17

example of reinforcement learning

Answer 17

skill acquisition, e.g. parked car simulator that has loads of trial and error

Question 18

process of training a machine for supervised learning

Answer 18

-define input database
-split the database into training and testing
-define the architecture of the network
-train the network
-validate/ test the network