Supervised Learning - Classification Week 9 Challenge
Answer 1. A neural network is a computational modelling spired by the way biological neural networks in the human brain work. It consists of interconnected nodes, or artificial neurons, organized in layers. Information is processed through these layers by adjusting the strengths of connections (weights) between neurons. Neural networks are used in machine learning for tasks such as pattern recognition, classification, regression, and other forms of data analysis.
Ans2.= Deep learning is a subset of machine learning that involves neural networks with multiple layers (deep neural networks). These networks, often referred to as deep neural networks, are capable of learning and representing complex hierarchical patterns in data. Deep learning has been particularly successful in tasks such as image and speech recognition, natural language processing, and various other domains where intricate patterns can be discovered through the automatic learning of hierarchical feature representations.
Answe 3.
The activation function in a neural network introduces non-linearity to the model by determining the output of a neuron. Without activation functions, the neural network would essentially be a linear model, unable to capture complex patterns and relationships in data. Activation functions enable neural networks to learn and represent non-linear mappings from input to output, allowing the network to model more intricate relationships and make the network capable of solving complex tasks, such as image and speech recognition. Common activation functions include sigmoid, hyperbolic tangent (tanh), and rectified linear unit (ReLU).
Answer 4.There are several types of activation functions used in neural networks. Here are some common ones:
Sigmoid Function (Logistic): Maps input values between 0 and 1. Often used in the output layer of binary classification models.
Hyperbolic Tangent Function (tanh): Similar to the sigmoid but maps input values between -1 and 1. It tends to be used in hidden layers.
Rectified Linear Unit (ReLU): Sets negative values to zero and passes positive values unchanged. It is widely used in hidden layers due to its simplicity and effectiveness.
Leaky Rectified Linear Unit (Leaky ReLU): Similar to ReLU but allows a small, non-zero gradient for negative values, preventing dead neurons.
Parametric Rectified Linear Unit (PReLU): An extension of Leaky ReLU where the slope for negative values is learned during training.
Exponential Linear Unit (ELU): A variation of ReLU that handles negative values more smoothly and introduces a slight negative slope for better learning.
Answer 5. Backpropagation, short for "backward propagation of errors," is a supervised learning algorithm used for training artificial neural networks. It is a method to optimize the weights of a neural network by minimizing the error between the predicted output and the actual target values.
The process involves the following steps:
Forward Pass: Input data is passed through the network, and predictions are calculated.
Calculate Error: The difference between the predicted output and the actual target values is computed, representing the error.
Backward Pass (Backpropagation): The error is propagated backward through the network. Partial derivatives of the error with respect to the weights are calculated using the chain rule of calculus.
Update Weights: The weights of the network are adjusted in the opposite direction of the gradient, reducing the error. This step is usually performed using optimization algorithms like gradient descent.
Repeat: Steps 1-4 are repeated iteratively until the network converges to a set of weights that minimize the error on the training data.