•Used machine learning techniques to simulate and eliminate printing defects, with a focus on neural networks.
•Experimented with architectures such as convolutional neural networks and generative adversarial networks.
•Built a dataset for subsequent use for training, evaluation, and analysis with machine learning models.
•Designed and experimented with loss functions based on frequency analysis and contrast sensitivity functions.
•Wrote code to implement a recurrent neural network-based version of the CycleGAN image-to-image translation model.