I had done a simulation study to see which model is better in prediction between ANN and SVM. ANN show good RMSE but high in bias. Is the result acceptable? What is the purpose of bias in simulation study? There are a lot researcher use bias in simulation.
Generally, bias in a Neural Network context refers to an always on node which outputs 1. It's use is similar to the intercept of a linear regression. Depending on your system, what shows up as 'bias' might just be the weight of the connection on this bias node. I.e. the interpretation of a large bias is similar to the interpretation of a large intercept. Whether this is an issue depends on the problem at hand but in prediction, it's probably not an issue.