Vicken Khachadourian

Processing Machine Learning Data without context is already having catastrophic and fatal consequences in daily life.

1 - A TESLA Autopilot or Full Self Driving is collecting a massive amount of data from multiple sensors and diverse sources. Then it evaluates that data and makes a decision on which way to continue its drive. Like Dr. Kennedy's analogy on my website if it mislabels the baseball game where the main attraction is the 10,000 hotdogs that got sold, or the 40,000 napkins that accompanied the hotdogs, then the TESLA will hit a parked Sheriff's car or a parked fire truck, all the time thinking it's got things right.

This is not an unconfirmed theory. Oracle conducted a 2.5 year investigation, proving conclusively that processing massive amounts of machine learning data in a short time, from diverse sources, without paying attention to context, will result in disaster.

2 - The same is true for when a driver is being monitored by sensors and cameras by fleet-owning companies. If a driver engages in a hard braking action because a kid was running after his soccer ball, he should not be penalized, but many sensors and in-dash cameras do not capture that context.



As an industry leading lab with the most important implementations for worldwide government and business, Oracle was warned about this problem in 1996, ignored it until 2007, does not have a solution today, because Oracle's trace files and diagnostics, that are in every module, were detected to be processing data out-of-context. Oracle can be implicated, sometimes criminally, for many of the failures in many mission critical systems, globally. Oracle's database Engine has been processing Machine Learning data out of context since 1996.


Vicken succeeded in detecting this problem on 400 of the toughest cases at Oracle Database Support over 13 years, with 100% success, solving problems in lightening speed and demonstrated the role of proper context for diags, but Oracle fired him for it. Larry Ellison stepped down in 2014 to focus on this problem and fix it, so far we are not seeing a solution. We're seeing meritless regulatory and litigation action by Oracle.