Lecture 1

Question 1

What is learning?

Answer 1

Learning is a process where a system improves performance from experience. Learns to do the task more efficiently the next time.

Question 2

Why learn? (3 examples)

Answer 2

• Developing systems that are too difficult/expensive to construt manually because they require specific detailed skills or knowledge tuned to a specific task
• Build software agents that can adapt to their users or to other software agents or to changing environments (autonomous systems, personalized tutoring)
• Discover new things or structures that were previously unknown to humans (data mining)

Question 3

What are some examples of tasks that are too hard to program?

Answer 3

• A self-driving car can’t be manually coded what to do and when, instead the car uses machine learning to know what to do and when based on the training data
• Autonomous ships: Sensors for situational awareness. Mapping, navigation and reactive collision avoidance. Satellites to know the state and communicate
• Optimizing sawmill production to maximize the yield of each log

Question 4

Is data cheap?

Answer 4

Data is cheap and abundant, knowledge is expensive and scarce

Question 5

What is machine learning?

Answer 5

Machine learning is programming computers to optimize a performance criterion using example data or past experience

Question 6

What are the components of machine learning?

Answer 6

• Unknown target function f, for example ideal credit approval function
• Hypothesis set: set of candidate formulas
• Training examples (data pairs x and y): historical records of credit customers
• Learning algorithm. Training examples and hypothesis set are given to the learning algorithm
• Final hypothesis: The learning algorithm provides function g that is close to the target function f.

Question 7

How is Artificial intelligence, Machine Learning and Deep Learning connected?

Answer 7

Artificial intelligence: A system that mimics human intelligence
Machine Learning: A subset of artificial intelligence that uses statistical techniques and algorithm to learn from data.
Deep learning: A subset of machine learning that consist of algorithm where in the system can train itself and take decisions