A majority of the cognative power of the human brain is devoted to the aquisition and processing of visual information. From the Vestibular systems which stablize our visual field, to bespoke neurons recognizing our mother's face. Visual information is the primary mode of human understanding. Our tools must understand, interperet, and communicate in our vision centred world.
Machine Learning, Deep Learning, A.I. All fascets of the same stone. As we ask more of our devices, as we gather more data, as we explore deeper, farther, smaller, and faster we need our tools to refine the deluge of information. Context, intent and interest must be understood on a case by case basis.
Computing power has traditionally been about relatively few physical things, doing the many different virtual things. Data only existed on our computers. This is changing with the rise of low power computer chips. Today's data exists in the real world, and a large number of physical things need to do a few virtual things very well.
Started by Dr. Steven Beauchemin The RoadLab Initiative is a NoAE Inovation Award winning "rolling labratory" designed to study how humans drive and to bridge the gap between human drivers and automation. This project introduced me to team coding and large scale software.
Entitled "Geological Object Recognition in Extraterrestrial Environments," my thesis was an exploration of how machine learning could be used to reduce bandwidth used by autonomous explorers by using iterative machine learning to allow a rover detect novelty within the scope of navagational imagery
Over the past few years, I have worked on side projects for different clients, as well as for my own edification and education. This has taught me a broad base of foundational skills for the IOT world. From efficency in code and memory use, to cross compilation and SQL data bases, to GIS intigration and RFID.
The real world is messy, and all the data processing in the world won't do you a lick of good if you don't get good, and relevant data. It's easy to focus on the clean numerical world of data processing, but if you're asking the wrong questions, your answers won't matter.
Scire Quod Sciendum