The NNML is currently involved in developing the following projects: a technique called "oracle-learning" which, as applied to artificial neural network learning, uses the most accurate artificial neural network available to label previously unlabeled data points to create a larger training set;I'm working on a way to detect overfit while training a neural network without requiring patterns withheld from training;classification training", a novel learning philosophy for backpropagation training of multi-layer neural networks on classification tasks; and several others