Neural computing'​s weight transport problem solved, enabling faster/ more accurate learning after pre-learning random information

https://www.eurekalert.org/news-releases/1068662

"biologically plausible error backpropagation learning possible without weight transport, overcoming slow learning speeds/ low accuracy, enabling use in neuromorphic computing... pre-trained with random noise, creating symmetry in forward/ backward neural cell connections, essential for error backpropagation learning... meta-learning how to learn... clues from developmental neuroscience, insight into the brain’s learning principles through artificial neural network models"

Related:

Laser-based artificial neuron mimics nerve cell functions at lightning speed
https://www.eurekalert.org/news-releases/1068365

Artificial protein-based neural network for cells classifies signals in cells
https://cen.acs.org/physical-chemistry/computational-chemistry/protein-based-neural-network-cells/102/web/2024/12