PersonAndrew Y. Ng / 吴恩达
Ng
A machine learning researcher and professor at Stanford University, known for the potential-based reward shaping theorem (1999) and contributions to online learning and deep learning.
Core Facts
Timeline (last 90 days)
Oct 14
Potential-based Reward Shaping was proposed by Ng et al. in 1999: if the intermediate reward is defined as F(s,s') = γΦ(s') - Φ(s), then the optimal policy will not change due to the shaping.
Unverified50%
All Facts (2)
Verified
Research by Ng et al. in 1999 proved that only potential-based reward shaping guarantees that the optimal policy remains unchanged.
65%UnverifiedPotential-based Reward Shaping was proposed by Ng et al. in 1999: if the intermediate reward is defined as F(s,s') = γΦ(s') - Φ(s), then the optimal policy will not change due to the shaping.
50%