Some issues about Socialway

[ChengzYuan]

Hi, there are some minor issues about your paper:

1. Performances in Socialways may not be reproduced;
2. Training and testing approach in Socialways different with the one in socialGAN paper (SocialGAN method: five subsets in total, ETH, ZARA1, ZARA2, HOTEL and UNIV; Using leave-one-out approach, train on 4 subsets and evaluating on the remaining set), so direct comparsion may not be reasonable;