Winner-take-all (WTA) networks can select the maximum from a set of data, so they are primarily used in decision making and selection. The Maxnet is a feedback WTA network. However, the Maxnet has two crucial problems. The first problem is its slow convergence rate. The second problem is that the Maxnet fails when non-unique maxima exist. In this work, dynamic inhibitory weights are used to speed up the convergence rate and a new convergence rule is proposed to enable the network to find all maxima. Simulation results indicate that the proposed network converges much faster than the other networks.