Expediting In-Network Federated Learning by Voting-Based Consensus Model Compression | IEEE Conference Publication | IEEE Xplore