The goal of this project is to build a simple but functional artificial neural network using only Numpy, a powerful library for numerical computations in Python. This will involve:
- Understanding the basic theory behind neural networks, including their architecture and how they learn.
- Implementing the neural network components such as layers, weights, biases, and activation functions.
- Designing a backpropagation algorithm to update the network weights based on error.
- Testing the network on simple datasets to observe its learning capabilitie
- Basic knowledge of Python programming.
- Familiarity with Numpy and its array operations.
- Understanding of basic calculus and linear algebra (mainly matrix operations).
- Basic concepts of machine learning such as training, testing, loss functions, etc.
- Implement the linear part of a layer's forward step, which is essentially z=wx+b (where w is weight, x is input, and b is bias).
- Apply an activation function like sigmoid, tanh, or ReLU to introduce non-linearity.
- Implement a loss function, such as mean squared error for regression or cross-entropy loss for classification, to evaluate the performance of the network.
- Calculate gradients of the loss function with respect to each weight and bias by applying the chain rule, moving backward from the output to the input layer.
- Update the weights and biases, typically using a simple gradient descent method or other optimization algorithms like SGD, Adam, etc.
- Combine the forward pass, loss computation, and backward pass into a training loop.
- Iterate over a set number of epochs or until the loss reaches an acceptable level.
- Optionally implement batch or mini-batch processing to improve training efficiency.
- Evaluate the trained model on a separate testing set to check its generalization capability.
- Fine-tune parameters or add complexity like more layers or different activation functions to improve accuracy
we used Dataset Breast Cancer
pip install -r requirements.txtFeel free to connect with me. Linkedin