news

한어Русский языкEnglishFrançaisIndonesianSanskrit日本語DeutschPortuguêsΕλληνικάespañolItalianoSuomalainenLatina

Chinese scientists build a new brain-like network to build a bridge between artificial intelligence and neuroscience

On the 16th, the reporter learned from the Institute of Automation of the Chinese Academy of Sciences that the team of researchers Li Guoqi and Xu Bo of the institute, in collaboration with Tsinghua University, Peking University and others, proposed a method for constructing brain-like neuron models "based on endogenous complexity" to improve the problem of computing resource consumption of traditional models and provide an example for the effective use of neuroscience to develop artificial intelligence. The relevant research was published in Nature Computational Science.

Building a more general artificial intelligence and enabling models to have more extensive and general cognitive capabilities is an important goal in the current development of the artificial intelligence field.

"The currently popular large-model path is to build larger, deeper and wider neural networks based on scaling laws, which can be called a general intelligence implementation method 'based on exogenous complexity'." Li Guoqi said that this path faces problems such as unsustainable computing resources and energy consumption and lack of explainability.

On the other hand, the human brain has 100 billion neurons and about 100 trillion synaptic connections. Each neuron has a rich and diverse internal structure, but the power consumption is only about 20 watts. Therefore, learning from the dynamic characteristics of brain neurons and enriching the neuron structure internally to explore general intelligence has great potential. This path can be called a general intelligence realization method "based on endogenous complexity."

Li Guoqi said that the experimental results verified the effectiveness and reliability of the endogenous complexity model in dealing with complex tasks, provided new methods and theoretical support for integrating the complex dynamic characteristics of neuroscience into artificial intelligence, and also provided feasible solutions for the optimization and performance improvement of artificial intelligence models in practical applications.

Currently, the research team has carried out further research, hoping to improve the computing efficiency and task processing capabilities of large models and achieve rapid implementation in actual application scenarios.

Reprinted from: Xinhua News Agency

Source: Workers Daily