Mixed Tabu machine for portfolio optimization problem

In this paper, we introduce a novel artificial neural network (NN) to solve the portfolio optimization problem. The proposed NN is called the Mixed Tabu Machine (MTM) since its structure is similar to the Tabu Machine, but includes both discrete and continuous variables. Similar to the Hopfield network, the state of the MTM is updated to find the global minimum energy state. To escape from local minimum states of the energy in the MTM, the state transition mechanism is controlled by a Tabu search in both discrete and continuous search spaces. The experimental results for five standard benchmark data sets show that the MTM can clearly obtain good results in very small computation time.