Timezone: »

 
Spotlight
Why Should I Trust You, Bellman? The Bellman Error is a Poor Replacement for Value Error
Scott Fujimoto · David Meger · Doina Precup · Ofir Nachum · Shixiang Gu

Thu Jul 21 12:50 PM -- 12:55 PM (PDT) @ Room 309
in Reinforcement Learning/Optimization »

In this work, we study the use of the Bellman equation as a surrogate objective for value prediction accuracy. While the Bellman equation is uniquely solved by the true value function over all state-action pairs, we find that the Bellman error (the difference between both sides of the equation) is a poor proxy for the accuracy of the value function. In particular, we show that (1) due to cancellations from both sides of the Bellman equation, the magnitude of the Bellman error is only weakly related to the distance to the true value function, even when considering all state-action pairs, and (2) in the finite data regime, the Bellman equation can be satisfied exactly by infinitely many suboptimal solutions. This means that the Bellman error can be minimized without improving the accuracy of the value function. We demonstrate these phenomena through a series of propositions, illustrative toy examples, and empirical analysis in standard benchmark domains.

Keywords: Reinforcement Learning · RL: Batch/Offline · RL: Deep RL · RL: Function Approximation

Author Information

Scott Fujimoto (McGill University)
David Meger (McGill University)
Doina Precup (McGill University / DeepMind)
Ofir Nachum (Google Brain)
Shixiang Gu (Google)

Related Events (a corresponding poster, oral, or spotlight)

More from the Same Authors