Machine Learning Quiz

Job ID: 32383947

Budget: $15 – $25 USD

a) Give the optimal Q-values for each action in each state and show an optimal policy, given: • A deterministic 4 state world (each cell is a state, depicted below)
• Top left has an intermediate reward of -1 for entering but is not an absorbing state
• Bottom left cell has a reward of 5 and is an absorbing state; no other rewards
• Only actions are move up, down, right and left, depending on the cell you are in. o e.g., the top left cell only has move right and down actions available
• The discount factor is .8
`
b) Give psuedocode for the general learning algorithm to learn approximate true Q values for arbitrary tasks.
Related categories: Statistics Machine Learning (ML)