Cascade correlation neural network with deterministic weight modification for predicting stock market price
Karuppanna Velusamy, R.V. Amalraj · IOP Conference Series Materials Science and Engineering · 2021
Abstract A new weight updating approach is proposed for cascade correlation neural network (CCNN) in this paper. The deterministic weight modification (DWM) algorithm is used to adjust the connection weights of CCNN. The introduced new method can improve the global convergence capability of the conventional CCNN and optimally reduced the system error. The proposed DWM+CCNN prediction algorithm is applied well-known stock market dataset in order to evaluate the robustness and efficiency. The experimental results are confirmed that the proposed DWM+CCNN algorithm is achieves higher performance in terms of convergence rate and the capability of global converges.