Feedforward neural network & other ML models implemented from scratch in pure C.
Models implemented:
- (multivariate) Linear regression. [linreg] [example]
- (multivariate) Polynomial regression. [polreg] [example]
- Artificial neuron (perceptron). [neuron] [example]
- Feedforward neural network. [nn] [example]
Examples are illustrated by fitting linear as well as non-linear functions.
For now the functions derivatives / weights gradients are calculated either "by hand" or approximated via finite difference for each example, but I plan on building an autograd engine soon too.
For the optimizer we always use the simple non-stochastic gradient descent algorithm.
For storing numerical values we use the single-precision floating-point type (float).