REG : Model Accuracy of 'Like Unlike Phone Classification' Problem

Sir,

I submitted the contest with an Accuracy of 68.9% (mp_neuron model).
But wanted to know how we can improve the MP_Neuron Model’s accuracy when there are no hyperparameters in the Model to tune, unlike the Perceptron Model - which has 2 hyperparameters, whose values can be tweaked to improve the accuracy.

Could you please provide some inputs on this?

Thank you.