Tests

Comparison with other neural networks

We have conducted extensive testing of PANN™ neural networks in comparison with other, classical, neural networks, using the popular IRIS and MNiST datasets (among others) on the same low spec hardware.

NB: Please note that the graphs below are truncated to be able to show PANN™ values visually. Otherwise, the enormous difference in performance makes PANN™ almost invisible in such graphs. See full details in the table below.

Comparison of PANN™ with Keras on TensorFlow by Google

Test details

Browse a summary of the tests carried out on PANN and classical neural networks.

Intellectual property

Join the pioneers in the AI revolution

Demo version

Specialists can use our demo version free of charge and check things out for themselves.

Dataset	Description	GitHub Repository	Computing time (Keras)	Computing time (PANN)	Times PANN is faster	Training Time (Keras)	Training Time (PANN)	Times PANN is quicker
Fashion MNIST	https://www.kaggle.com/qnkhuat/fashionmnist	https://github.com/Hlaiman/p-net-comparisons	2.5740	0.0572	45	74,279	3,333	22
Pima Indians diabetes	https://www.kaggle.com/uciml/pima-indians-diabetes-database	https://github.com/Hlaiman/p-net-comparisons	0.5277	0.0041	127	549	12	46
Wine quality	https://archive.ics.uci.edu/ml/datasets/Wine+Quality	https://github.com/Hlaiman/p-net-comparisons	0.5151	0.0051	101	3,812	28	136
Banknote authentication	https://archive.ics.uci.edu/ml/datasets/banknote+authentication	https://github.com/Hlaiman/p-net-comparisons	0.5068	0.0039	127	5,030	17	296
Combined cycle power plant	https://archive.ics.uci.edu/ml/datasets/combined+cycle+power+plant	https://github.com/Hlaiman/p-net-comparisons	0.8255	0.0035	234	4,426	4	1,107
Boston house price	https://www.kaggle.com/vikrishnan/boston-house-prices	https://github.com/Hlaiman/p-net-comparisons	0.6557	0.0036	183	689	1	689
Abalone	https://archive.ics.uci.edu/ml/datasets/abalone	https://github.com/Hlaiman/p-net-comparisons	0.8386	0.0035	239	3,752	3	1,251