From the course: Deep Learning: Model Optimization and Tuning

An artificial neural network (ANN) model

From the course: Deep Learning: Model Optimization and Tuning

Start my 1-month free trial Buy for my team

An artificial neural network (ANN) model

“

- [Instructor] What is a neural network model and how do we build it? A neural network model is represented by a set of parameters and hyperparameters. The parameters include the weights of weights and biases for all the nodes. The hyperparameters includes a number of levers, like layers, notes in a layer, activation functions, cost functions, learning rate, and optimizers. Training an ANN model means determining the right values for these parameters and hyperparameters, such that it maximizes the accuracy of predictions for the given use case. In this course, we will look at ways to optimize these hyperparameters to achieve the best results. How do we train a model? We use training data, like regular machine learning, where we know both the dependent and independent variables. We will start with a network architecture by intuition. We also initialize weights and biases to random values. Then we repeat the iterations of applying weights and biases to the inputs and computing the error. Based on the error found, we will adjust the weights and biases to reduce the error. We keep repeating the process of adjusting weights and biases until the error gets to an acceptable value. We will also fine tune the network hyperparameters to improve training speed and reduce iterations. Finally, we will save the model as represented by its parameters and hyperparameters, and use it for predictions.

Contents