FiRe 2017: Pictorial Processing Beyond the Exascale

By Ahrash Poursaid

Larry Smarr and Mark Anderson talking about pictorial supercomputing

Reverse engineering the human brain to further delve into the understanding of its functionality and the human psyche is a complex task and seems impossible but we are close to accomplishing it, according to Larry Smarr, founding director of California Institute for Telecommunications and Information Technology (Calit2).

“The brain is our biologically evolved computer systems”, Smarr said during a panel at the Future in Review (FiRe) conference in Park City, Utah. He said that to mimic it, exascale computing is required, which is 1 million teraflops faster than a smartphone.

“Exascale simulates 100% of the complexity of the human brain in real-time”, said Smarr. He elaborated the capabilities of machine learning and to track the neural networks within the brain.

“With the incredible speed of digital computers, a human being cannot keep up with this prodigious flow of data.” said Smarr.

He emphasized that for the human brain to keep up the digital information, it needs to be put through a visualization device such as an oscilloscope for the eye-brain system will be able to interface with the natural world.

The advances in the field have allowed AI to learn voice recognition. The use of Tensor Processing Units (TPU) has become an integral part of this world and provides the capability for machines to learn and recognize the different voices of a family and sync the calendar accordingly to each member.