Freelance data scientist · All case studies

A line, then a net, both written by hand.

An iterative perceptron for a linear decision boundary, then a shallow network with custom backpropagation and SGD — tanh and linear activations, error you can plot, weights you can print.

Client: Shallow-net study. Built by Dilshad Raza.

The linear case had to show the iterative update until the boundary sat between the classes. The shallow net had to own its backward pass — tanh in the hidden layer, linear where the head needed it — and SGD that actually descended the error you plotted.

I implemented an iterative perceptron that resolves a linear recommendation boundary, then wrote a shallow network with custom backpropagation and stochastic gradient descent using tanh and linear activations — error curves included.

Hire Dilshad Raza for similar freelance data science, machine learning, and AI automation in the UK, United States, and Australia.