three-layer-neural-network

In this project, the multilayer artificial neuralnetwork algorithm implemented with python language. The project supports 2 output and 3 output networks.

Calculate Loss

Cross-entropy loss applied.

Predict

tanh and softmax activation functions used.

Build Model

This function takes 2D input array and calculates 2D prediction vector. While building a model forward- propagation performed for making a prediction, and backpropagation applied to update the network parameters.

Build Model 691

This function takes 2 input and classify 3 output labels.

Results

2D Input & 2D Output
I used Moon dataset from sklearn.
X, y = make_moons(200, noise=0.20)

2D Input & 3D Output
I used Moon isotropic Gaussian blobs dataset from sklearn.
X, y = make_blobs(n_samples =100, centers=3, n_features=2, random_state=0)

This project developed for the Machine Learning Course (CS 691) at UNR.