T+ D- Normal size
In an effort to create a system for speech recognition and computations, scientists connected brain tissue to electronic devices; With the aim of making significant progress in research to develop high-performance biological computing devices.
The work is inspiring efforts to develop so-called neural computing devices that work with human brain cells, learn faster and are more energy efficient compared to traditional silicon-based devices. As the researchers say, progress in this emerging field could lead to advanced applications of artificial intelligence, leading to major breakthroughs in fields such as medical and therapeutic sciences.
“Brain-inspired computing devices aim to mimic the structural and functional principles of the brain,” the authors of a paper on this hybrid innovation published in the journal Nature Electronics.
Researchers from Indiana University Bloomington, Lureda University, University of Cincinnati and Cincinnati Children’s Hospital Medical Center developed the brainware system using so-called brain organoids, three-dimensional neural structures grown from human stem cells produced in the laboratory.
The team wired computers to send electrical stimuli to nerve cells and study the neural activity they produced in response. The system was able to recognize Japanese vowel sounds and predict mathematical functions.
The scientists tasked the Brainware language testing system with distinguishing between eight different Japanese-speaking men recorded from a set of 240 audio clips. The system improved significantly after training its basic algorithms; Its accuracy increased from about 51% to about 78%.
In a math test, the researchers tried to get the computer to predict the Henon diagram, a representation of a random function. Here, Brainware’s system proved to be slightly less accurate than silicon-based neural networks, but its training time was over 90%.
The research comes after an international team led by scientists at Johns Hopkins University in the US launched a comprehensive map of “neurointelligence” in February. The program promises innovation in other areas of computing, neuroscience and medical research.
Although the researchers acknowledge that general-purpose biological computers may be decades away from being implemented, Brainware’s experiments highlight the field’s potential.
As three Johns Hopkins scientists wrote in a commentary published in the journal Nature Electronics, biocomputing research has the potential to generate “fundamental insights into learning, neurodevelopment, and the cognitive consequences of neurodegenerative diseases.” “In the next few years, more complex neural systems that can interact with increasingly complex artificial environments are likely to emerge,” said three researchers who were not involved in the BrainWare paper.
On the other hand, this progress raises ethical questions about the creation of “intelligence on a plate” similar to the brain and the potential to acquire basic consciousness. “As the complexity of these neural organoid systems increases, it will be necessary for the community to consider the myriad of neurobiological issues surrounding biological computing systems that include human neural tissue,” the researchers added.
“Creator. Award-winning problem solver. Music evangelist. Incurable introvert.”
