• Artificial intelligence researchers have celebrated successes with neural networks, but they remain relatively inflexible.
• In 2020, two researchers at MIT introduced a new kind of neural network based on the *Caenorhabditis elegans* worm.
• Liquid neural networks offer an elegant and compact alternative, and experiments are showing they can run faster and more accurately than other continuous-time neural networks.
• Liquid networks differ in how they treat synapses, and they are more adaptable than traditional neural networks.
• The team has tested the network on an autonomous car and an aerial drone, and they are working to improve the network’s architecture.
• Liquid networks are well suited to the analysis of electric power grids, financial transactions, weather, and other phenomena that fluctuate over time.
Published February 10, 2023
Visit Nautilus to read Steve Nadis’s original post Researchers Discover a More Flexible Approach to Machine Learning
