Automatic combinational circuit generation with neural networks

Master thesis

In this master thesis, an automatic combinational circuit generation with neural networks will be explored. Specifically, a specific test bench is first provided as the training data for a neural network. Afterwards, the neural network is trained with the data while maximizing the accuracy. After training, the neural networks will be quantized and thus converted to a combinational logic design according to the following figure. The candidate also needs to evaluate the effectiveness of this automatic combinational circuit generation method.