Chapter 3 - Supervised Learning (BPNN) Flashcards
What is the core idea of backpropagation in a neural network?
The core idea is to adjust the weights in each layer of the network based on the error, similar to how gradient descent is used in linear regression
What type of relationships can a Backpropagation Neural Network learn?
It can learn non-linear relationships through multiple hidden layers and activation functions
What are the three main phases of the Backpropagation Neural Network?
The three main phases are feedforward, backpropagation of the error, and weight changes
How does backpropagation help in training a neural network?
It helps calculate the gradient of a loss function with respect to all the weights in the network. Also, it fine-tunes the weights based on the error rate obtained in the previous epoch
What is a key advantage of using a Backpropagation Neural Network?
It is fast, simple, and easy to program. It is also flexible as it does not require prior knowledge about the network and does not need any special mention of the features of the function to be learned
What are some limitations of linear regression that backpropagation addresses?
Linear regression can only model linear relationships and is limited to a single output. Backpropagation can model non-linear relationships
How does a perceptron compare to a Backpropagation Neural Network?
A perceptron is a simple neural network that can only solve linearly separable problems. A Backpropagation Neural Network can learn non-linear relationships.
What does it mean to “backpropagate the error?”
It refers to the backward propagation of errors, and is a method of fine-tuning the weights of a neural net based on the error rate obtained in the previous epoch
What is one of the main advantages of using backpropagation when training a neural network?
Proper tuning of the weights allows you to reduce error rates and to make the model reliable by increasing its generalization.
What is one disadvantage of using BPNN?
The actual performance of backpropagation on a specific problem is dependent on the input data. Backpropagation can also be quite sensitive to noisy data.
Why is backpropagation useful for deep neural networks in areas like image or speech recognition?
Backpropagation is especially useful for deep neural networks working on error-prone projects, such as image or speech recognition.
What is the purpose of studying the group of input and activation values in backpropagation?
To develop the relationship between the input and hidden unit layers.
According to Fausett (1990), how many hidden layers are generally sufficient in a BPNN?
One hidden layer is generally sufficient, although more than one may be beneficial for some applications.
What is the relationship between linear regression, perceptrons and backpropagation?
Backpropagation is an extension of both linear regression and perceptrons.
