Supervised Learning - Classification Week 9 Challenge

1. A neural network is a computational model composed of interconnected nodes organized into layers, designed to mimic the human brain's structure. It processes information by transmitting signals through the network, enabling learning from input data and making predictions or classifications.

2. Deep learning is a subset of machine learning that employs neural networks with multiple hidden layers. It facilitates the automatic learning of hierarchical representations from data, enabling the extraction of intricate features and patterns, often leading to state-of-the-art performance in various tasks.

3. The activation function introduces non-linearity to neural networks, enabling them to learn complex relationships within data. It determines whether a neuron should be activated or not, influencing the network's output. Activation functions are crucial for enabling neural networks to approximate complex functions and make accurate predictions.

4. Common activation functions include:

Sigmoid: S-shaped curve, useful for binary classification tasks.

Tanh (Hyperbolic tangent): Similar to sigmoid but centered at zero, aiding in faster convergence.

ReLU (Rectified Linear Unit): Piecewise linear function, computationally efficient and effective in overcoming vanishing gradient problems.

Leaky ReLU: Variant of ReLU with a small gradient for negative inputs, preventing dead neurons.

Softmax: Converts raw scores into probabilities for multiclass classification tasks.

5. Backpropagation is a key algorithm for training neural networks. It adjusts the network's weights iteratively by propagating error gradients backward from the output layer to the input layer. This process enables efficient learning from labeled training data, updating the model's parameters to minimize prediction errors and improve performance.