I am working on a Shallow Neural Network that can be used to read input configurations for Cloud Computing Environments and determine whether it is energy efficient. I have turned the measurements in my dataset I am using into quantile values to make it easier for my network to read. However, after running it I keep getting 55% accuracy and a loss function of about .68. I am new to building Neural Networks, so I am certain there are some things I can do to increase accuracy. I have tried many different methods to see if I can get a value above 55%, but to no avail. I tried adding more layers, adjusting the pixel values, converting my variables from their original continuous values to quantiles values, but nothing can make it above 55%. I'm uncertain of whether the dataset is the reason this will not go above this number, but any advice would help, thank you! Neural Network Design Continue reading...
