This paves the way for brain-mimicking computing, helping establish artificial intelligence

This paves the way for brain-mimicking computing, helping establish artificial intelligence

Artificial intelligence and machine learning have the potential to truly revolutionize how the world works today, but their development is hampered by the fact that the current state of the art in electronics is not aligned with what is needed. hindered.

In a move toward developing devices that can mimic the workings of neurons in the brain, researchers at the Indian Institute of Science (IISc) in Bengaluru are using a previously unused organic material to create neuromoles. designed a fic device. Their work since 2014 has been aimed at this.

Organic materials were considered the poorest of the various types of materials when it comes to creating computing components, as they are fragile and unstable. “I chose this genre as my racing horse because I believed that if there was a way to solve these performance issues, the features that could be extracted from these materials would blow away anything that existed,” says Center. says Sreetosh Goswami of for. Nanoscience and Engineering (CeNSE), IISc.

He and his collaborators have published significant papers in the field since 2017. Natural materials, nanotechnology in nature, advanced materials When Nature, which proves that organic materials can be reliably calculated and are superior to inorganic materials in some aspects. “The molecular system (transition metal complexes of azoaromatic ligands) was the brainchild of my father, Professor Sreebrata Goswami,” Dr. Sreetosh Goswami said in an email. hindu.

plastic brain

In the words of Sreebrata Goswami, now at CeNSE, IISc, the human brain that has inspired the researchers’ work is “in terms of its ability to learn, perceive, and make decisions, it resembles an artificial electronic analogue. It’s far better than.” Its amazing performance draws just 20 watts of power in a 1260 cc space. Some of the desirable properties it exhibits include interconnectivity and reconfigurability.

“Neurons in the brain operate on the brink of chaos with highly nonlinear feedback mechanisms. We are looking for materials that can capture such properties, but this is an elusive goal.” …” explains Professor Sreebrata Goswami.

many features

Molecular materials present a multidimensional landscape of parameter space that is characterized by interactions between molecules and ions and can be tuned to develop appropriate functions.The question they asked in a recent paper published in advanced materials The question was whether these many-body interactions could be manipulated to achieve device plasticity and reconfigurability. They did this by measuring current-voltage curves as a function of temperature over a wide range. They can capture features across bipolar, unipolar, nonvolatile, and volatile memristors.

In the words of Dr. Sreetosh Goswami, this is an “abnormal amount of variability”, and to explain it, the group uses mathematical techniques that allow almost any possible characteristic variation desired in a neuromorphic device. I had to design the space. “The same device can operate in both analog and digital modes, just by adjusting the activation energy,” said Dr. Sreetosh Goswami.

make it work

A challenge has been that during low temperature measurements, molecular memristors exhibit quenching or flattening of the switching response as the temperature decreases. “We were able to make it work because our molecular device is robust and the switching mechanism has a thermodynamic component that occurs even when the device is cooled,” said CeNSE, IISc, and First author of a paper published in advanced materials.

Dr. Sreetosh Goswami said:


Source link


Submit a Comment

Your email address will not be published. Required fields are marked *