Triple-GAIL: A Multi-Modal Imitation Learning Framework with Generative Adversarial Nets

Authors: Cong Fei, Bin Wang, Yuzheng Zhuang, Zongzhang Zhang, Jianye Hao, Hongbo Zhang, Xuewu Ji, Wulong Liu

IJCAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental Experiments on real driver trajectories and real-time strategy game datasets demonstrate that Triple-GAIL can better fit multi-modal behaviors close to the demonstrators and outperforms stateof-the-art methods.
Researcher Affiliation Collaboration Cong Fei1,2 , Bin Wang1 , Yuzheng Zhuang1 , Zongzhang Zhang3 , Jianye Hao1 , Hongbo Zhang1 , Xuewu Ji2 and Wulong Liu1 1Huawei Noah s Ark Lab 2Tsinghua University 3Nanjing University
Pseudocode Yes Algorithm 1 The Training Procedure of Triple-GAIL
Open Source Code No The paper does not provide any explicit statements or links indicating that the source code for the described methodology is open-source or publicly available.
Open Datasets Yes The experiment is conducted with the Next Generation Simulation program s Interstate 80 (NGSIM I-80) dataset[Halkias and Colyar, 2006].
Dataset Splits No The paper mentions sampling 150 trajectories for the driving task and evaluating performance metrics, but it does not specify explicit numerical splits for training, validation, and testing datasets (e.g., percentages or exact counts for each split).
Hardware Specification No The paper does not provide any specific details about the hardware (e.g., GPU models, CPU types, memory) used to run the experiments.
Software Dependencies No The paper does not provide specific version numbers for any software dependencies, libraries, or programming languages used.
Experiment Setup Yes For the generator network, the images are input to a 3 layer convolutional network with 64 128 256 feature maps, while the vectors are run through fully connected network with 256 256 hidden units with a final layer expands its size the same as the output of the convolutional network. The skill label is run through fully connected network with 128 128 hidden units with corresponding expansion to the size of input states. Then the input states and the skill label are now the same size and concatenated together, and are run through fully connected network with 256 128 hidden units to output actions and value. The selector network has the same architecture of the generator with input states and actions then outputs skill label. The discriminator network is similar to the selector, and we just adjust all hidden units to 64 for the input states, actions and skill label, then run through fully connected network with 128 64 hidden units to output reward value. In the RTS task, the state information includes current resource of players and the images with 20 20 dimensions.