Diffusion learning over adaptive competing networks