EasyDeepLearn

Reward clipping in DQN — trade-offs.

medium

Answer

  • DQN Atari clips rewards to {-1, 0, +1} → stabilizes Q-learning across games with vastly different reward scales (Ms. Pac-Man vs Skiing).
  • Cost: agent doesn't distinguish 1-point vs 100-point reward → suboptimal for scale-sensitive tasks.
  • Modern replacement: reward normalization by running SD (used in PPO), or distributional RL that models absolute return distribution.
  • Best of both worlds: keep scale info but normalize for gradients.
Check yourself — multiple choice
  • Always needed
  • Clip to {-1, 0, +1} stabilizes across scales but loses magnitude info; modern replacement: reward normalization by running SD (PPO) or distributional RL absolute returns
  • Random
  • Not needed

Reward clip: stable but lossy; PPO uses running-SD reward normalization.

#deep-rl#engineering

Practise Reinforcement Learning

214 interview questions in this topic.

Related questions