Generative adversarial imitation learning-based solution(G-RML) Different from traditional communication solutions, the source encoder of the proposed G-RML does not focus only on sending as much of the useful messages as possible; but, instead, it tries to guide the destination user to learn a reasoning mechanism to map any observed explicit semantics to the corresponding implicit semantics characterized by a set of possible reasoning paths involving the hidden entities and relations that are most relevant to the semantic meaning. By applying G-RML, we prove that the destination user can accurately imitate the reasoning process of the source user and automatically generate the reasoning paths following the same probability distribution as the expert paths. Compared to the traditional semantic communication solutions, our proposed G-RML requires much less communication and computational resources and scales well to the scenarios involving the communication of complex semantic meanings with a large number of concepts and relations.

pip install -r requirements.txt

download the dataset and unzip in ./SemanticInference FB15K237 NELL-995

in ./SemanticEncoder

# run
python train_transe --dim 200 --gpu 1 --dataset FB15K237

in ./SemanticInference

# run
python sl_policy.py $relation
python policy_agent.py $relation retrain
python policy_agent.py $relation test
# or you can run the shell script
./pathfinder.sh ${relation_name}

# run 
python fact_prediction_eval.py $relation
./link_prediction_eval.sh ${relation_name}

Yong Xiao, Yiwei Liao, Yingyu Li, Guangming Shi, H. Vincent Poor, Walid Saad, Merouane Debbah, and Mehdi Bennis, "Reasoning over the Air: A Reasoning-based Implicit Semantic-Aware Communication Framework," accepted at IEEE Transactions on Wireless Communications (TWC).