Name two desiderata besides crash avoidance that a helicopter reward function must trade off.
Question: Besides avoiding crashes, name two other factors the reward function R(.) must trade off according to the source.
Sample answer: Two other factors are how bumpy or smooth the landing was and whether the helicopter landed at the exact desired spot; ride roughness for passengers is another valid factor.
Key points:
- Landing smoothness
- Landing at the desired spot
- Ride roughness during descent
Rubric: Full credit for naming any two of: landing smoothness, landing location accuracy, ride roughness; partial credit for one correct factor.
0
1
References
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Machine Learning Yearning (Deeplearning.ai)
Tags
Data Science
Foundations of Large Language Models Course
Computing Sciences
Machine Learning
Deep Learning
Supervised Learning
Dive into Deep Learning @ D2L
Machine Learning Strategy
Machine Learning Yearning @ DeepLearning.AI
Related
In the helicopter reinforcement learning example, what does the reward function R(T) primarily measure?
True or False: A trajectory ending in a helicopter crash could be assigned a reward like R(T) = -1,000.
A safe landing trajectory T might result in a _____ R(T), with the exact value depending on landing smoothness.
Match each trajectory outcome in the helicopter example to its typical reward value.
Order the steps for designing and using a reward function in the helicopter RL example.
Analyze why designing a good reward function for helicopter trajectories is difficult.
Diagnose why a helicopter RL agent keeps producing unsafe landings despite high average reward.
Name two desiderata besides crash avoidance that a helicopter reward function must trade off.
Who typically chooses the reward function R(.) in the helicopter reinforcement learning example?
True or False: The reward function only needs to account for whether the helicopter crashes, ignoring passenger comfort.