AI Paper Atlas

Research map / Reinforcement learning / Visual and representation RL

Phase-Parametric Policies for Reinforcement Learning in Cyclic Environments

Arjun Sharma, Kris Kitani

AAAI 2018 · Reinforcement learning › Visual and representation RL · approach: Inferring decision-making from behavior · 7 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add phase-parametric layers instead of phase as inputDRQN · Flying with Wind pwind=1.0return: -13.7 -> -2.9Improved
add phase-parametric layers instead of phase as inputDRQN · Flying with Wind pwind=0.0return: 2.0 -> -1.6Hurt
add phase-parametric layers instead of phase as inputDQN · Flying with Wind pwind=0.6return: 0.4 -> 0.5No clear effect
add phase-parametric layers instead of phase as inputDQN · Flying with Wind pwind=0.0return: 2.4 -> -0.5Hurt
add phase-parametric layers instead of phase as inputDDPG · Hopper short trainingaverage return: 2133.34 -> 1397.72Hurt
train longer (40000 vs 10000 episodes)Phase-DDPG · Hopperaverage return: 1397.72 -> 2630.16Improved
train longer (40000 vs 10000 episodes)DDPG-IP · Hopperaverage return: 2133.34 -> 2559.47Improved
train longer (40000 vs 10000 episodes)Phase-DDPG · Walkeraverage return: 2532.24 -> 4491.60Improved
initialize control weights from a DDPG-NP model then fine-tune (ablation)Phase-DDPG · Hopperaverage return: 1397.72 -> 3325.57Improved
initialize control weights from a DDPG-NP model then fine-tune (ablation)Phase-DDPG · Walkeraverage return: 2532.24 -> 5124.44Improved
use feed-forward Phase-DQN instead of recurrent Phase-DRQNPhase-Parametric Q-network · Freeway On-Ramp R4:8,0:0return: 0 -> 8Improved
vary wind probability pwind from 0.0 to 1.0Phase-DRQN · Flying with Windreturn: -1.6 -> -2.9No clear effect

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy