The IceCube neutrino detector, located in the clear glacial ice at the South Pole, completed construction in 2011. The low-energy infill extension, DeepCore, forms a denser sub-detector using higher quantum efficiency photosensors. DeepCore has been taking data since May 2010 and lowers IceCube's energy threshold to about 10 GeV. These low-energy events are dim compared to higher energy events, necessitating the study of low-light backgrounds. While Monte Carlo predictions give an expected rate of approximately 6 Hz due to atmospheric muons, DeepCore records a significantly higher rate of 13.5 Hz with most of the discrepancy due to unsimulated noise events. Much of the rate difference may be resolved by rejecting especially dim events by counting the number of locally coincident hits, retaining 55% of the νe and 62% of the νμ events sampled with energies of 10-300 GeV while rejecting 96% of noise events. However, differences in the timing distributions of noise hits indicates a need for further study. A new source of correlated noise has been discovered, necessitating an updated noise simulation model. IceCube's new noise generator is able to reproduce the correlated noise in both IceCube and DeepCore sensors. A Metropolis-Hastings algorithm has been used to identify relevant parameters for nearly all of the 5160 sensors that make up the IceCube detector. Initial low quality fits reduce the rate discrepancy between data and simulation from 32% using a Poissonian noise model to 20% using the updated noise model with additional reduction possible. Noise which triggers the DeepCore detector is evaluated and rejected using the NoiseEngine filtering module. Minimal cleaning removes 94% of noise triggers while retaining 85% of νe and 87% of νμ events with energies of 10-300 GeV. Stringent cleaning removes 99.9% of noise triggers while retaining only 55% of νe and 60% of νμ signal events in the same energy range.