How much can neural networks learn about jellyfish?

Abstract

Neural Networks can approximate real-valued, discrete values and vector-valued functions. Artificial neural networks (ANNs) have inputs neurons, hidden layers, and output neurons. A Convolutional Neural Network (CNN) has additional steps before the ANN: convolutional filtering and pooling stages. Hyperparameters are parameters that impact CNN performance: e.g., filter size, pooling type, number of hidden layers, number of hidden layer neurons, batch size, and number of iterations. The research questions for this study are: 1) Can a network learn what is a jellyfish vs what is a worm? and 2) Can we predict jellyfish attributes from vortex wakes?