Brain research could help scientists create new kinds of AI. Here’s how:

Sometimes nature does the best when it comes to computers too.

MIT researchers claim to have solved the puzzle behind the interaction of two neurons. This unlocks a new class of fast artificial intelligence (AI) algorithms. It’s part of a growing research area where brain research can help create advanced new forms of AI.

“Brain research is aimed at understanding the communication between individual neurons or groups of neurons, which helps us understand natural intelligence,” says Carnegie Mellon University’s Mechanical Engineering Associates, who was not involved in the MIT research. Engineering professor Rahul Panat told Lifewire. email interview. “Natural intelligence can then be used to develop artificial intelligence.”

think better

An MIT team created a neural network that outperformed its state-of-the-art counterparts on a multitude of tasks, resulting in significant speed and performance gains in recognizing human activity from motion sensors, and improved performance of simulated walking robots. We modeled the physical dynamics.new paper nature machine intelligenceFor example, the new model was 220 times faster at sampling 8,000 patients on the medical prediction task.

“A new machine learning model called ‘CfC’ replaces the differential equations that define neuron computation with closed-form approximations, preserving the beautiful properties of liquid networks without the need for numerical integration,” said senior at MIT. Professor Daniela Rus said. The author of the new paper said in her news release: “CfC models are causal, compact, explainable, and efficient to train and predict. They pave the way for reliable machine learning for safety-critical applications.”

Brain-inspired computer advances are already available. Reinforcement learning (RL) is a popular family of AI algorithms inspired by recent advances in brain research used in self-driving cars and robotics, says a professor of bioengineering at the University of California, Riverside. ‘s Vasileios Christopoulos said in an email. interview.

“RL is an adaptive process in which a species uses previous experiences in order to use them to make recent advances. We were able to develop an algorithm in which autonomous agents learn to behave in an unfamiliar environment by performing actions and observing the results of their actions,” he added. rice field.

think about the future

Panat predicts that improved recording will eventually allow scientists to interpret and decipher communications between different parts of the brain.

“The nervous system that controls the idiosyncratic behavior of animals is important for AI,” he added. “For example, we are studying how neurons in octopuses communicate and coordinate their movements to complete specific tasks.”

But deciphering what is happening inside the human skull is a daunting task. Panat pointed out that the brain has about 80 billion neurons.

“Investigating communication between different parts of the brain at the cellular level is a very challenging task,” he said. “We are currently limited to recording up to 1000 neurons talking to each other via neural probes, so advances in recording density and interpretation of neuronal signaling patterns are of great interest.”

As researchers learn more about the similarities between silicon and neurons, computer technology may one day help treat brain disorders, too. His Maryam Ravan, a professor of electrical and computer engineering at the New York Institute of Technology, recently co-authored a study using machine learning to improve treatments for mental health conditions. In one study, scientists analyzed a patient’s brain waves and developed a machine-learning algorithm to classify the patterns as biomarkers for bipolar disorder or major depressive disorder.

“Interest in AI technology is growing as society becomes more accepting and open about mental health,” Ravan said in an email interview. We hope to continue to see additional research leveraging forms of AI, including machine learning, that can help streamline the treatment and diagnosis of cancer.”