Rubric

Keep in mind that 25 students have already been assessed using this rubric. Changing it will affect their evaluations.
A4
A4
Criteria Ratings Pts
Exercise 1
Kernel Matrix and GP Regression
threshold: pts
40 pts Exceeds (1) Proving PSD of Kernel Matrix (10 pts) (2) Feature Map for Specific Points (10 pts) (3) GP Regression Implementation (20 pts)
0 pts No Evidence
pts
40 pts
--
Exercise 2
Game Theory - Two-Player Game (20 pts)
threshold: pts
20 pts Exceeds (a) Nash Equilibrium for R (10 pts) (b) Iterated Deletion (10 pts)
0 pts No Evidence
pts
20 pts
--
Exercise 3
Driving game
threshold: pts
30 pts Exceeds (a) Payoff Matrix (3 pts) (b) Rationalizable Strategies (3 pts) (c) Pure Strategy Nash Equilibria (4 pts) (d) Mixed NE: Row Pure, Column Mixed (5 pts) (e) Mixed NE: Both Mix Between Two (7 pts) (f) Fully Mixed NE (8 pts)
0 pts No Evidence
pts
30 pts
--
Exercise 4.1
Inverted Pendulum via MDP: MDP Construction
threshold: pts
20 pts Exceeds Discretizes state and control space (4 pts) Correctly wraps x1 , handles grid edge cases (4 pts) Computes deterministic next state (3 pts) Approximates Gaussian transitions (5 pts) Cost function implemented correctly with hℓ(x,u) (4 pts)
0 pts No Evidence
pts
20 pts
--
Exercise 4.2
Solving the MDP
threshold: pts
20 pts Exceeds Implements either value or policy iteration (6 pts) Visualizes value function convergence (4 pts) Compares both methods for convergence speed and behavior (6 pts) Provides meaningful written analysis (4 pts)
pts
20 pts
--