The simulation shows a grid of binary neurons whose pairwise weights are set by Hebbian learning from a handful of stored patterns, visualizing the network's energy landscape and watching a noisy or partial cue update neuron by neuron, monotonically descending in energy until it locks onto the nearest stored memory.
Choose which patterns to memorize, then corrupt or partially erase a cue pattern and press play to watch asynchronous updates flip neurons one at a time; track the energy value as it descends step by step, and try adding more patterns to see when recall starts failing near the ~0.14N capacity limit.
Pattern selection, noise/corruption slider, single-step and play/pause update controls, energy readout, network size and stored-pattern-count adjustment
John Hopfield's 1982 paper on these networks helped launch the modern resurgence of neural network research, and in 2024 he was awarded the Nobel Prize in Physics for this foundational work connecting statistical physics to associative memory and machine learning.
The simulation shows a grid of binary neurons whose pairwise weights are set by Hebbian learning from a handful of stored patterns, visualizing the network's energy landscape and watching a noisy or partial cue update neuron by neuron, monotonically descending in energy until it locks onto the nearest stored memory.
The simulation shows a grid of binary neurons whose pairwise weights are set by Hebbian learning from a handful of stored patterns, visualizing the network's energy landscape and watching a noisy or partial cue update neuron by neuron, monotonically descending in energy until it locks onto the nearest stored memory.
Choose which patterns to memorize, then corrupt or partially erase a cue pattern and press play to watch asynchronous updates flip neurons one at a time; track the energy value as it descends step by step, and try adding more patterns to see when recall starts failing near the ~0.14N capacity limit.
John Hopfield's 1982 paper on these networks helped launch the modern resurgence of neural network research, and in 2024 he was awarded the Nobel Prize in Physics for this foundational work connecting statistical physics to associative memory and machine learning.