A lot of science fiction predicts that machines will eventually start acting like people. Even before scientists and engineers start to work on them, there has long been talk about machines that can learn, think, and make judgments like people.
Although not much progress has been made yet, the world of artificial intelligence and machine learning is expected to make many exciting improvements.
In order to let artificial intelligence (AI) continuously evolve over time, researchers at Purdue University in the US are creating AI hardware that is inspired by the human brain.
They want to make it possible for autonomous vehicles or space robots to use AI in the future. The benefit of directly embedding AI into hardware as opposed to running it in software is that the machines on which it is embedded will be more efficient.
Engineers at MIT have created a brain-like chip that is smaller than confetti and has thousands of artificial brain synapses, or memristors, on a single chip.
Memristors, also known as memory transistors, are silicon components that resemble the information-transmitting synapses in the human brain. One day, it's possible that these so-called "brains on a chip" may be integrated into compact, portable devices, enabling them to do complicated computations that are currently only achievable on supercomputers.
The researchers at the Northwestern University and the University of Hong Kong have developed a device inspired by human brain that can learn like humans. This device learns by association with the help of synaptic transistors. Synaptic transistors are capable of processing and storing information at the same time.
Following are the details of the three projects which are aimed at making the computers capable of imitating the human brains:
The Purdue Project
A paper was published in Science in February 2022, in which Developers at Purdue University explained how computer chips can keep themselves active and able to accept new data like a brain, enabling AI to continuously learn.
The brain is constantly making connections between neurons to continue the learning process. Similarly, to create a brain-like computer chip, the circuits on top of the chip have to be changed. However, the circuit that the computer has been using for years is the same circuit that was built for the computer at the factory.
According to Shriram Ramanath, a professor at Purdue University, researchers should be able to program, reprogram or replace the chip. Ramnath and his team created new hardware that can be reprogrammed with electrical signals as needed.
The team's thinking is that because the device is adaptable, it will be able to perform all the functions necessary to create a computer inspired by the human brain.
According to Purdue; "By simulating experimental data, Purdue teammates at Santa Clara University and Portland State University demonstrated that the device's internal physics creates a dynamic framework for an artificial neural network that interprets electrocardiogram patterns and digits is able to be identified more efficiently than a static network.”
This neural network makes use of "reservoir computing" which tells that how different parts of the brain communicate with each other and transmit information.
The team now wants to demonstrate this on large-scale test chips that could be used to create a brain-like computer, the researchers said.
The Project "Brain-on-a-chip" From The MIT
According to a statement;
"MIT researchers are working toward the day when people can carry artificial brains (that work), untethered to the Internet, supercomputers, or cloud devices."
"Like a brain synapse, a memristor will be able to 'remember' the value associated with a given current, and the next time it is applied," the statement said. A similar current will produce exactly the same signal."
In an article on the Nature NanoTechnology website, the scientists explain how their brain-like computer chip memorized the image of Captain America's shield and set each pixel on the Memristor exactly as it was in the real image, and thus reproduced the image of the shield many times over.
In this article, MIT researchers highlight the fact that this "brain-on-a-chip" can be used to perform complex tasks on mobile devices, tasks that currently only supercomputers can handle.
Brain-like Device That Mimics The Human Brain's Learning Process
In a study published in Hong Kong, researchers from Northwestern and the University of Hong Kong, imitating Ivan Pavlov, a Russian psychologist who conditioned dogs to receive food by ringing a bell, trained their computing device with light and pressure.
According to this research, the device mimics the human brain by using electrochemical "synaptic transistors", which are capable of processing and storing information at the same time. These synapses make the brain highly parallel, error-tolerant, and enable energy efficient operation.
The plastic synapses of this device work exactly like natural biological synapses.
With its brain-like potential, the unique transistors and circuits could potentially overcome the limitations of traditional computing, including their energy-sapping hardware and limited ability to perform multiple tasks at once.
This brain-like device is also highly fault tolerant, continuing to function smoothly even when some components fail. Current computer systems keep memory and logic separate, but combining the two together would save space and reduce energy costs.
And the new computing device's soft, plastic-like polymers will enable researchers to integrate it into smart robotics, wearable electronics, and even devices implanted in people.
Conclusion
Although the ubiquity of robots is not yet fully realized, these improvements are evidence that scientists are working to replicate the brain's biological machinery, and that such computers may one day will be able to work like a human brain.
Read More:
What Is 3D Printing And What Are Its Uses?
