Wednesday, November 30, 2016

Using Scavenger Hunts To Solve Real World Problems?

In a recent article in the news site "Research & Discover Magazine" titled "DARPA Completes 2nd Field Test of Experimental Nuclear Detection" a scavenger hunt was used to test a network of sensors.

This is wonderful for the enhancement of defense for the country's biggest challenges. With over a 1000 sensors deployed, the challenge is to fuse all of the data together to make sense of the efficiency and challenges that lie ahead in the real-time detection of a threat.  The amount of data coming in from such an exercise (simulation) must be overwhelming.  Further, to process and correlate the data is time consuming.

In a recent post I wrote about the NFL collecting 'big data', a single football game generates a large data set.  The number of sensors (one each players shoulder pad + sensors on the field) should generate a data set of considerable length.  The authors did not specify the size of either data set in each study -- which begs the following questions:

How does the data set generated by a single football game compare the data set generated by the scavenger hunt above?

I wonder if the two organizations are using similar algorithms to mine the data sets generated out of each circumstance (scavenger hunt vs. football game)?

The two different studies obviously generate different amounts of data.  Furthermore, the types of sensors used in each study are probably quite different.  Regardless, each of these studies prove that the future holds a considerable investment into generating large data sets.  Additionally, new graduates of the physical and life sciences would be well-served to have the experience in handling large data sets.

The future is exciting for those working in data science.  As in other fields, as time goes by and problems are tackled, new problems emerge.  If the problems are not tackled, then no new problems will emerge.  In the case of generating and handling big data, correlations might be possible but a lack of computational power might be limiting.

More often the case, a researcher is stuck with a large data set and lacks the questions regarding the possible mining strategies.  In order to come up with possible correlations, scientist must understand the data that is being collected in each study (scavenger hunt vs. football game).   Although, with the help of mining data from social media companies, progress could be spread up exponentially.  This necessitates the need to have more interdisciplinary research.  Otherwise we have a bunch of studies occurring in parallel which might benefit if the researchers would talk to each other about the problems and strategies without giving away proprietary information.

This means that we will be at a point of stagnation -- which is not good.   Fortunately, most scientists find a way to move forward and tackle problems in order to keep pushing the field of research forward.  At the current moment, both DARPA and private organizations like the NFL do not appear to be running out of funding, which is exciting for those of us watching the developments from the 'sidelines.'

No comments:

Post a Comment