Deep learning
Signal rasterization
Computer vision
Genetics
Simulation studies
Signal processing
Availability:
Unrestricted
Abstract:
Statistical machine learning approaches are quite famous for processing Markov signal data. They can model unobserved states and learn certain characteristics particular to a signal with good accuracy. However, with the advent of Deep learning the novice ways of solving a problem has shifted towards this more sophisticated algorithm, which is much better, powerful and more accurate. Specifically, Convolutional Neural Nets (CNN) have shown many promising results on images and videos. Here we illustrate how CNN can be applied to a 1D numeric signal using signal rasterization technique. We start by rasterizing a 1D numeric Markov signal into an image followed by applying CNN to perform two basic tasks: signal classification and error localization. We call this process as RM-Net. We demonstrate the performance of our approach on simulated data benchmarked against baselined statistical models. We also illustrate the supremacy of our technique on real word dataset 1000 Genomes Project Phase 3 SV where we try to estimate the location of Copy Number Variant (CNV) in a chromosome. Finally, we conclude using the metrics obtained on both the datasets that our proposed approach is much better, shows promising results and has scope for future improvements over traditional statistical machine learning approaches.
If you have any questions please contact the ETD Team, libetd@njit.edu.