This project investigates how Multiple Constant Multiplication (MCM) could be used to accelerate the computations of Neural Networks. The compute resource will be used for MCM optimization solvers. GPUs will also be needed for train and evaluate the inference of benchmark models.