This research project presented itself to me with the potential to strengthening my Data Science and being able to dive myself further into academia and research before graduating. Before truly getting tangled in the research, I knew I would thoroughly enjoy learning about Machine Learning and Convolutional Neural Networks with its interdisciplinary utility. As this research focuses on using Computer Vision for image analysis of bruises, but fundamentally using feature detection can be executed on just about anything of interest.
With the long-term goals, I want to fortify my background in Computational Data Sciences with cutting edge machine learning techniques that will always be in my back pocket for Civil Engineering purposes. This isn’t replacing engineering work or design, only re-enforcing better techniques and computations. In terms of data fusion, this can be exceedingly supportive with decision making confidence especially when it is relying on data-founded understanding and prediction. Technology is going to ultimately make everybody’s life easier in the long run, and I would like to be part of the solution; if that is in the medical/health informatics field, Civil Engineering field, or any other.
The week begins with a Monday morning group meeting discussing our successes and failures and how to improve our trajectory for the research and well as key ideas to explore. During the week we are mostly conducting in depth research and experimentation; this is mostly writing scripts, troubleshooting, organizing input data and possibly running neural network trials, comparing and analyzing results. The week concludes with a team meeting on Friday with our other partner research groups, Health Informatics and Nursing, along with all respective faculty mentors discussing our goals and polishing our methodology. During my experience I was exploring the vast power of Convolutional Neural Networks to the benefit of the user. As a quick synopsis Convolutional Neural Networks is using a multitude of weighted equations to match many complicated features to a specific output. Not knowing how profoundly powerful different types of network architectures can be when placed in the right situation