MIT launches Neuromorphic Chip that can work like a Human Brain

Researchers in the field of neuromorphic computing at the Massachusetts Institute of Technology (MIT) are trying to design neuromorphic chips that function like the human brain.

First, let's see how the human brain works. Neurons in our brain signal signals to other neurons through synapses (intercellular spaces). The synapse functions as a bridge for neuronal communication. There are about 100 billion neurons in the brain, each neuron releases 5 to 50 messages per second. This activity will allow us to deal with our environment, move our muscles, make conscious movements, make decisions and form memories.

A team of researchers at the Massachusetts Institute of Technology attempted to design an artificial synapse that precisely controls the intensity of the current flowing through it, just like an ion flows between brain nerve cells.

Researchers have created a neuromorphic chip consisting of artificial synapses made of silicon germanium. Each size is about 25 nanometers. They applied a voltage to each synapse and discovered that all synapses showed about the same current or ionic current and about 4% difference between synapses. The introduction of this "brain on" chip works in analog mode (unlike computer chips used to process digital signals), so we can handle multiple parallel calculations like our brain .

The team tested a single synapse in multiple tests and applied the same voltage over 700 cycles and found that the synapse exhibited the same current with only 1% change between cycle and cycle. In the simulation, they discovered that chips and their synapses can be used to identify handwritten samples with a 95% correct rate.

Jin, who is responsible for MIT's research team, says:

It can prove to be a major advance in portable artificial intelligence devices (smart phones, laptops etc.). Using these neuromorphic chips makes it easier to use devices that use AI technology. This makes it faster and more efficient and consumes less power.

"Intel's Neuromorphic Computing Team is creating a chip that simulates how the human brain observes, learns and understands. The company's prototype chip" Loihi "is the latest step in this direction . Please explain the data. (Credit: Intel) "In principle, there is no reason why the tracking camera can not track the position of the controller and headset at the same time, so there is no reason to install the 3dof controller in a 6dof Daydream headset with built - in tracking capability Microsoft The Hybrid Reality headset tracks the position of the headphones, the controller is 6 dof, there are two tracking sensors inside and outside Oculus Santa Cruz prototype also did this, I tried this because of technical limitations Rather than the imagination limit, does this include the ARcore delivery function?

Then there is the neuromorphic computing project of the European human brain project. It is working on building two large and unique neuromorphic machines and prototyping the next generation of neuromorphic chips. Recently, a paper demonstrating usefulness in computational neuroscience applications was released by completely simulating 80,000 neurons based on SpiNNaker hardware and 300 million synaptic cortical microcircuit models. Co-author of this paper, Professor Markus Diesmann, Director of Computational Systems Neurosciences at the Jülich Research Center in Germany, says: "There is a big gap between brain energy consumption and today's supercomputer.How much can you achieve energy efficiency of the brain using electronic equipment?"