Unsupervised Video Object Segmentation for Deep Reinforcement Learning

Authors: Vikash Goel, Jameson Weng, Pascal Poupart

NeurIPS 2018 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Sec. 5 evaluates the approach empirically on 59 Atari games. Finally, Sec. 6 concludes the paper and discusses possible future extensions.
Researcher Affiliation Academia Vik Goel, Jameson Weng, Pascal Poupart Cheriton School of Computer Science, Waterloo AI Institute, University of Waterloo, Canada Vector Institute, Toronto, Canada {v5goel,jj2weng,ppoupart}@uwaterloo.ca
Pseudocode No The paper does not contain any structured pseudocode or algorithm blocks.
Open Source Code Yes Our code is available at https://github.com/vik-goel/MOREL.
Open Datasets Yes We showcase the performance of MOREL on all 59 Atari games where we observe a notable improvement in comparison to A2C and PPO for 26 and 25 games respectively, and a worse performance for 3 and 9 games respectively.
Dataset Splits No The paper mentions training on Atari games and conducting an ablation study but does not explicitly provide details about training, validation, or test dataset splits.
Hardware Specification No The paper mentions the use of 'Cry SP RIPPLE Facility at the University of Waterloo' but does not provide specific hardware details such as GPU/CPU models or memory amounts used for experiments.
Software Dependencies No The paper mentions optimizers and algorithms but does not provide specific software dependencies with version numbers (e.g., Python, PyTorch, TensorFlow versions).
Experiment Setup Yes We collect 100k frames by following a random policy. Using an Adam optimizer [18] with learning rate 1 × 10−4 and batch size 16, we minimize Ltotal for 250k steps. Following the experimental setup from [24], we train each agent for 10 million timesteps with one timestep for each frame.