Convolutional neural networks are typically used to perform face and radar recognition, object recognition, and signal classification.

This network was trained on the CIFAR-10 dataset (Learning Multiple Layers of Features from Tiny Images, Alex Krizhevsky, 2009) comprised of 50,000 training images and 10,000 testing images. Each image is classified as an airplane, automobile, bird, cat, deer, dog, frog, horse, ship, or truck.

This network was configured with these layers and hyperparameters. After 16 epochs of training, it reached an accuracy of 86%. That accuracy is good enough considering that this is just a demo, and it is 8.6 times more accurate than making a random guess.

Try it out! Select a testing image that the network has never seen before and we'll show you its prediction of the image's classification.

Perceptron neural networks are typically used to give computers the ability to search data for patterns and trends, then turn those findings into business insights and predictions.

This network was trained on the Iris dataset (R. A. Fisher, 1938) comprised of 150 plants with their sepal length, sepal width, petal length, and petal width. Each plant is classified as Iris setosa, Iris versicolor, or Iris virginica. We used 120 plants (40 from each classification) as the training plants, and 30 plants (10 from each classification) as the testing plants.

This network was configured with these layers and hyperparameters. After 458 epochs of training, it reached an accuracy of 100%.

Try it out! Select a testing plant that the network has never seen before and we'll show you its prediction of the plant's classification.