Erica Sherkin
Contributing Writer
Kumar Vinjeet, a third-year Ph.D. student at UC Santa Barbara, is currently working on a project to build a synaptic device that mimics brain-like neural networks. He aims to create a form of higher energy computing with lower energy consumption, which could lead to faster computer processing, more efficient use of artificial intelligence (AI) without significant battery drainage, or even create a roadmap that helps us better understand the processing in our brains themselves.
“Our brain does so much processing [just] based on the food we eat, [and] all of these processes take place so fast and efficiently,” Vinjeet told The Bottom Line (TBL). In other words, the goal for most researchers in this field is to mimic the processing that takes place in our brains, so they can use that processing on a material scale.
An umbrella term for Vinjeet’s general research is called Photonics, which analyzes light waves and energy. Vinjeet, however, is more specifically working on neuromorphic computing, or neuron-like computing, which tries to replicate how energy is transferred in the brain. This research is not necessarily limited to building hardware, but it also includes programming and designing electrical circuits.
In describing the device he is building, Vinjeet explained, “the material is the most important part of this device.” The researchers pass electrical signals through a constant material, changing the intensity of light and wavelengths. These electrical signals manipulate the material itself, specifically the conductivity of the material. Vinjeet can determine if the device took on a “short-term memory change” if the material resets after the light or signal is turned off, or a “long-term memory change” if it remembers their conductivity change.
There are a wide variety of applications for new neuromorphic computing developments, from driverless cars to domestic robots. Neuromorphic technology can unlock energy-efficient cognitive functions in many artificial intelligence technologies, and companies like the International Business Machines Corporation are invested in cognitive computing as their “main business for the future”.
Additionally, Vinjeet told TBL that neuromorphic computing is trying to get as close to a biological approach as possible, mimicking human and animal-like synapses that could supercharge these connections. This level of processing is more energy efficient, and it could enhance already existing AI services, as well as become the main hardware for the processing in driverless cars.
While this area of research is relatively new, Vinjeet explained that both his research advisor, as well as his group of peers, have been nothing short of helpful on his project. Excited to continue his studies, Vinjeet encourages anyone who has a “curious brain” to consider jumping into research opportunities on campus.
“Research is like any other field … [it is another way] to get good at something. If you are a person that is [constantly] trying to understand the how’s and why’s of something, I would encourage you to go into research.”