In June Nutonian, along with a number of leaders from national research facilities and companies, was invited to present to Congress to showcase the scientific and technical advancements happening across the country. This conference was organized by the National User Facility Organization (NUFO), which represents national research facilities open to the public.
As the event was focused on scientific innovation, we took the opportunity to show off how our machine intelligence software can be used to significantly improve the underlying performance of characterization equipment – cameras, medical imaging equipment, scanning electron microscopes, etc. – at the heart of many medical and scientific breakthroughs.
Instead of just looking for the common image artifacts or distortion present in imaging equipment (i.e., fish eye lens), we targeted the harder questions: 1) Does changing the imaging equipment settings change your answer? 2) Can Eureqa identify the underlying relationships governing how the image is being distorted? 3) Can we use these relationships to significantly improve the accuracy of measurements made with the equipment? Working with both x-ray image data from the Advanced Photon Source at Argonne National Laboratory and scanning electron microscopy data, we showed that the answers to these questions are yes, yes, and absolutely (see Tuesday’s blog post for more detail). Our findings showed that for the systems considered, we were able to reduce the measurement error by >10x.
We had captivating conversations with real decision-makers like Dr. Patricia Dehmer, Deputy Director of Science Programs for the Department of Energy, and a few Representatives, most notably a former entrepreneur, then physicist, then Congressman Representative Bill Foster from Illinois. All discussions started at the distortion tool we were showing off, but quickly moved on to our vision for the future: automated data science lowering the barriers for anyone leveraging complex analysis to gain understanding from their data. This could help dramatically improve a variety of circumstances today, ranging from more targeted, lower-cost healthcare treatment to energy efficiency to crime reduction. The biggest impediment to making the world a “smarter planet” is our inability to quickly discern what the waterfalls of data actually mean and which factors actually matter. Flow that data through Eureqa, however, and the most important elements will automatically come to light.
As a society, what if we could isolate the true drivers of obesity? What if we could distill the most effective compounds to treat harmful illnesses, become more precise about electricity generation, or predict and eradicate a famine – before locals even knew it was imminent? Problems like the distortion issue we presented often go unsolved, and worse, unidentified. By harnessing the power of machine intelligence, we can accelerate the next wave of discovery, not only for businesses, but for people. Happy modeling.