Requirements for different grades

Grade E

For grade E the student should be able to use the terminology and solve standard problems in machine learning. In particular, the student should be able to:

• Define and correctly use standard machine learning terms
• Use Find-S and Candidate Elimination on standard problems
• Use feedforward ANNs for linear and non-linear classification
• Design and use decision trees
• Use bayesian classification on simple problems
• Use temporal difference algorithms on standard problems
• Use a k-nearest classifyer

Grade D

For grade D the student should also be able to manually do the calculations of important machine learning algorithms and be aware of the limitations. In particular:

• Calculate predictability of data using entropy
• Describe how boosting works and when it can be used
• Calculate the weights of ANNs from a set of data
• Define overtraining and suggest methods to avoid it

Grade C

For grade C the student should also be able to apply the algorthms to novel problems.

• Apply an evolutionary algorithm to a novel problem
• Define a game or other problem as a reinforcement learning task
• Use an ANN to solve real-world classification and function approximation problems

Grade B

For grade B the student should also be able to use the theory.

• Define and use the conecpt of PAC-learnability
• Define and use the concept of VC dimension
• Apply the Bellman equation to a RL task

Grade A

For grade A the student should be able to use and describe all the techniques brought up during the course with no or very few errors.