Highway Graph to Accelerate Reinforcement Learning
Abstract: Reinforcement Learning (RL) algorithms often suffer from low training efficiency.
A strategy to mitigate this issue is to incorporate a model-based planning algorithm, such as Monte Carlo Tree Search (MCTS) or Value Iteration (VI), into the environmental model.
The major limitation of VI is the need to iterate over
a large tensor with the shape $|\mathcal{S}|\times |\mathcal{A}| \times |\mathcal{S}|$, where $\mathcal{S}/\mathcal{A}$ denotes the state/action space.
This process iteratively updates the value of the preceding state $s_{t-1}$ based on the state $s_t$ in one step via value propagation. These still lead to intensive computations.
We focus on improving the training efficiency of RL algorithms by improving the efficiency of the value learning process.
For the deterministic environments with discrete state and action spaces, on the sampled empirical state-transition graph, a non-branching sequence of transitions can directly bring the agent from $s_0$ to $s_T$ without deviating from intermediate states, which we call a \textit{highway}.
On such non-branching highways, the value-updating process can be merged as a one-step process instead of iterating the value step-by-step.
Based on this observation, we propose a novel graph structure, named \textit{highway graph}, to model the state transition.
Our highway graph compresses the transition model into a concise graph, where edges can represent multiple state transitions to support value propagation across multiple time steps in each iteration.
We thus can obtain a more efficient value learning approach by facilitating the VI algorithm on highway graphs.
By integrating the highway graph into RL (as a model-based off-policy RL method), the RL training can be remarkably accelerated in the early stages (within 1 million frames).
Moreover, a deep neural network-based agent is trained using the highway graph, resulting in better generalization and lower storage costs.
Comparison against various baselines on four categories of environments reveals that our method outperforms both representative and novel model-free and model-based RL algorithms, demonstrating 10 to more than 150 times more efficiency while maintaining an equal or superior expected return, as confirmed by carefully conducted analyses.
Submission Length: Long submission (more than 12 pages of main content)
Changes Since Last Submission: N/A
Assigned Action Editor: ~Steven_Stenberg_Hansen1
Submission Number: 2667
Loading