Hopfield network implemented with Python.
- Python3 for _train.py
- Python2.7 for example.py
- numpy
- matplotlib
- tqdm
- PIL
- cv2
Run _train.py.
The following is the result of using Asynchronous update.
Start to data preprocessing...
Start to train weights...
100%|██████████| 4/4 [00:06<00:00, 1.67s/it]
Start to predict...
100%|██████████| 4/4 [00:02<00:00, 1.80it/s]
Show prediction results...
or
or
Show network weights matrix...
Is an application for drawing.
For now run PyPaint/example.py
Is to make this two app, become one, the Hopfield Network would be on background of PyPaint only print your results.
The PyPaint will have a signal that call a slot, from time to time, that saves an image of drawing. That image will send to HN via thread execution that will have an answer to draw.
- Amari, "Neural theory of association and concept-formation", SI. Biol. Cybernetics (1977) 26: 175. https://doi.org/10.1007/BF00365229
- D. Pebly, "PyQt Painting widget", PyQtPaint (2017): https://github.com/dpebly/pyqt-paint
- J. J. Hopfield, "Neural networks and physical systems with emergent collective computational abilities", Proceedings of the National Academy of Sciences of the USA, vol. 79 no. 8 pp. 2554–2558, April 1982.