Föreläsning 3 (Intro till datavetenskap)

Question 1

What are 6 examples of things rule-based AI is good at?

Answer 1

-Path-finding (paths in trees for example, or navigation systems)
-Games
-Control systems
-System verification
-Theorem proving
-Proof verification

Question 2

When is rule-based AI preferred over machine learning?

Answer 2

When data plays no role, and there are rules to follow.

Question 3

What is machine learning?

Answer 3

A computer program learns from experience E with respect to some
task T and some performance measure P, if its performance on T, as
measured by P, improves with experience E.

ML is the study of algorithms that learn based on data.

Question 4

What are the 3 types of ML?

Answer 4

-Unsupervised learning
-Supervised learning
-Reinforcement learning

Question 5

For what types of problems is unsupervised learning used?

Answer 5

• Dimentionality reduction (big data visualisation etc)
  -Clustering (recommendor systems, targeted marketing, customer segmentation)

Question 6

For what types of problems is supervised learning used?

Answer 6

-Classification (customer retention, image classification, diagnostics, identity theft detection)
-Regression (advertising popularity prediction, waether forecasting, population growth prediction)

Question 7

Match the right type of ML to the following:
1. Data without labels -> ____ -> mapping: data to classes
2. States, actions, rewards -> _____ -> mapping: states to actions
3. Data with labels -> _____-> mapping: data to labels

Answer 7

1. Unsupervised learning
2. Reingforecement learning
3. Supervised learning

Question 8

Name 4 data sources!

Answer 8

-Artistic works
-Social media
-Traditional media
-Scientific data

Question 9

Describe the data science pipeline in 6 steps.

Answer 9

1. Collect data
2. Clean data
3. Visualize data
4. Build model
5. Evaluate model
6. Use model

Question 10

Name the 6 ways of visualising datasets as mentioned in the lecture

Answer 10

-Histogram
-Scatter plot
-Box plot
-Violin plot
-Quiver plot
-Heat map
(look up which is good for what type of data if you’re unsure!)