Lecture 10 - Part 2: Machine Learning

Question 1

What can human intelligence do?

Answer 1

• Solve problems
• Achieve goals
• Analyse & reason
• Communicate
• Collaborate & influence
• Consciousness
• Emotions, Intuition, Imagination

Question 2

What can artifical intelligence do?

Answer 2

The ability for machines to simulate & enhance (human) intelligence.

Question 3

Define AI

Answer 3

The designing and building of intelligent agents that receive precepts from the environment and take actions that affect that environment.

Question 4

What is machine learning?

Answer 4

Machine Learning is a subfield of Artificial Intelligence focused on developing algorithms that learn to solve problems by analysing data for patterns.

Question 5

What is deep learning?

Answer 5

Deep Learning is a type of Machine Learning that leverages Neural Networks and Big Data

Question 6

What are some applications of machine learning?

Answer 6

• Pictures (make difference between objects)
• Voice to text

Question 7

Image classifcation
Object identification
Segmentation
Machine translation
Reccomendations

Answer 7

REFER TO LECTURES

Question 8

What is the machine learning paradigm?

Answer 8

Rules based learning to get answer (REFER TO SLIDES)
- Example: activity recognition (in slides) (REFER TO LECTURE REWATCH PART)

Question 9

What is the basic process of machine learning?

Answer 9

Make a guess -> Measure accuracy -> Optimise your guess (REFER TO SLIDES FOR EXAMPLE)

Question 10

What is the loss function - Mean Square Error

Answer 10

REFER TO SLIDES

Question 11

What are the categories of machine learning?

Answer 11

Supervised learning -> made up of: regression and classification
Unsupervised learning -> made up of: clustering, segmentation, etc

Question 12

What is reinforcement learning?

Answer 12

• The agent has a task to perform
• It takes some actions in the environment
• It gets feedback telling it how well it did on performing the task
• The agent gets positive reinforcement for tasks done well
• The agent gets negative reinforcement for tasks done poorly

Question 13

What are AI enablers and its drawbacks?

Answer 13

• Data availability - drawback: privacy and security
• Computational power - drawback: large use of power
• Algorithm advancements - drawback: data bias, likes something more than the other, more of something was used to train making it biased
• Broad public interest - drawback: AI is the solution for everything which is not the case