Class Nine Flashcards
What is a neural network?
A neural network is a computational model inspired by the structure and functioning of the human brain. It consists of interconnected nodes, called neurons or units, that work together to process and transmit information.
What are the advantages of neural networks?
Advantages of neural networks include their ability to learn and generalize from large and complex datasets, their capability to handle nonlinear relationships, and their suitability for tasks like pattern recognition and prediction.
What are Threshold Logic Units (TLUs)?
Threshold Logic Units (TLUs) are basic computational units in neural networks that produce a binary output based on a threshold function applied to the weighted sum of their inputs. They can be used as building blocks for more complex network architectures.
What are the limitations of Threshold Logic Units (TLUs)?
Limitations of TLUs include their inability to model complex decision boundaries or capture fine-grained distinctions, as they only produce binary outputs.
What is a Percepton?
Perceptron is a simple neural network with a single layer of nodes, primarily used for binary classification.
What are the disavantages of Perceptons?
- Perceptron incapable of learning complex patterns:
- Because decision boundary of each output neuron is linear (just like Logistic Regression classifiers).
- Incapable of solving Exclusive OR (XOR) classification problem
- Soln: Can be solved by stacking multiple Perceptrons
- Resulting ANN is called a Multilayer Perceptron (MLP).
What is a Multi-Layer Perceptron (MLP) with many layers?
A Multi-Layer Perceptron with many layers refers to a neural network architecture that consists of multiple layers of interconnected neurons, including input, hidden, and output layers. It allows for complex representations and learning of hierarchical features.
What are the advantages of a Multi-Layer Perceptron with many layers?
Advantages of an MLP with many layers include its ability to learn and represent complex relationships in data, its flexibility in capturing both simple and complex patterns, and its potential for achieving high levels of accuracy.
What are the limitations of a Multi-Layer Perceptron with many layers?
Limitations of an MLP with many layers include the potential for overfitting if the model becomes too complex or the training data is limited, the challenges in interpreting the learned representations, and the increased computational complexity.
What is a feedforward network?
A feedforward network is a type of neural network where information flows in one direction, from the input layer through one or more hidden layers to the output layer, without any cycles or feedback connections. It is often used for tasks such as classification and regression.
What are the advantages of a feedforward network?
Advantages of a feedforward network include its simplicity and ease of implementation, the availability of efficient training algorithms (e.g., backpropagation), and its ability to approximate complex functions and learn from large datasets.
What are the limitations of a feedforward network?
Limitations of a feedforward network include the requirement for labeled training data, the potential for overfitting if the model becomes too complex, and its inability to handle sequential or temporal data without additional techniques (e.g., recurrent neural networks).
How does training a neural network occur?
Training a neural network involves an iterative process where the model’s weights and biases are adjusted based on the error between its predictions and the true labels. This is typically done using optimization algorithms like backpropagation to update the parameters.
What is backpropagation?
Backpropagation is an algorithm used to train neural networks by computing the gradients of the loss function with respect to the model’s parameters. It propagates the error backward through the network, allowing for parameter updates that minimize the error.
What is the activation function in a neural network?
An activation function determines the output of a neuron or a layer in a neural network based on the weighted sum of its inputs. Common activation functions include sigmoid, ReLU, and tanh, which introduce nonlinearity and enable the model to learn complex relationships.
(More: MaxOut, Leaky ReLu, ELU)
