How artificial neural networks build their own learning machinery

MIT News reports that in a paper published Oct. 9 in the journal PNAS, scientists at MIT's McGovern Institute for Brain Research and K. Lisa Yang Integrative Computational Neuroscience (ICoN) Center[1] argue that a network's ability to organise itself is a key measure of whether its learning is biologically plausible.According to MIT News, the team found that with some simple instructions about how and when connections should change, a group of randomly connected units can reshape itself into an effective information-processing network[1].Co-first author Qianli Liao told MIT News that "Current AI technology is working very well, but it doesn't necessarily use a learning algorithm that is used by the brain."[1]MIT News reports that most modern AI implements gradient descent through some form of the back-propagation algorithm, which neuroscientists say is unlikely to be used by the brain[1]. According to the outlet, the team compared its self-organizing network to networks designed to learn in this way and found that it performed at a similar level[1].According to Physical AI News, which republished the MIT story, the team showed that such a network can learn to recognize patterns and classify images[2].Professor Tomaso Poggio told MIT News, as republished by Mirage News, that if the brain uses the proposed motifs, "we will have a strong connection at the level of fundamental mechanisms between artificial and natural networks"[3].Tech Xplore reports that Liao points to the development of brain-computer interfaces[4] as one possible use.