Hebbian models Unsupervised models where learning takes place with no feed-back from the environment and with redundancy in the input data. Implemented learning rules: Sanger, Oja and plain Hebbian.


Hebb net and learning phase: the initial configuration is represented by the clouds of green dots, eventually the principal components are found (3 components and Sanger rule in this case).

Video Video of a Hebb net computing the two principal components of a 3D Gaussian distribution
Image Compression A tool to compress images, that implements a Hebbian algorithm.
Next model