Block 1: Fundamentals of Machine Learning

Question 1

Name one application of neural networks mentioned in the text.

Answer 1

Handwriting recognition.

Question 2

What is the primary function of the sigmoid output function in neural networks?

Answer 2

The sigmoid function is used to map the output of a neuron to a value between 0 and 1.

Question 3

What are neural networks typically used for in machine learning?

Answer 3

They are used for learning, modeling, and making predictions based on data.

Question 4

What is the significance of weights in a neural network?

Answer 4

Weights determine the strength of the signal that neurons send to each other, impacting the network’s overall output.

Question 5

Define the term ‘overfitting’ in the context of neural networks.

Answer 5

Overfitting occurs when a model learns the training data too well, including its noise and fluctuations, and performs poorly on new, unseen data.

Question 6

What is the role of the loss function in neural networks?

Answer 6

The loss function measures how well the neural network is performing, guiding adjustments to improve accuracy.

Question 7

Discuss the ethical considerations of using neural networks.

Answer 7

Ethical considerations include ensuring fairness, transparency, and accountability in decision-making processes, and addressing privacy and data security concerns.

Question 8

How does a three-layer network learn through errors?

Answer 8

It learns by adjusting its weights based on the errors in the output compared to the expected result, a process known as backpropagation.

Question 9

What are the challenges in initialising weights for datasets in neural networks?

Answer 9

Challenges include avoiding too large or too small weights, which can lead to slow learning or the vanishing gradient problem, respectively.

Question 10

Explain how the error function is used in training neural networks.

Answer 10

The error function measures the difference between the network’s predicted output and the actual output, guiding the adjustment of weights to minimize this error.

Question 11

Discuss the trade-offs in improving neural network outputs, considering data size and applications of weights.

Answer 11

Trade-offs include balancing model complexity with the risk of overfitting, managing computational resources, and ensuring the model’s generalizability to new data.

Question 12

Analyze the potential limitations of neural networks in terms of overfitting and interpretability.

Answer 12

Overfitting limits a network’s ability to generalize from the training data to new data. Interpretability issues arise from the ‘black box’ nature of neural networks, making it hard to understand how decisions are made.

Question 13

Detail the process of backpropagation in neural networks using a worked example.

Answer 13

Backpropagation involves computing the gradient of the loss function and propagating this information back through the network to update weights. This helps the network learn from its errors.

Question 14

Explore the implications of the vanishing gradient problem in neural network training.

Answer 14

The vanishing gradient problem occurs when gradients become too small, drastically slowing down training or stopping it altogether, particularly in deep networks.

Question 15

Analyze the role of training parameters in determining the success of a neural network model.

Answer 15

Training parameters like learning rate, batch size, and the number of epochs significantly influence the model’s learning efficiency and accuracy.

Question 16

Assess the ethical implications of applying neural networks in real-world scenarios.

Answer 16

Ethical implications include potential biases in decision-making, privacy concerns, and the need for transparent and accountable AI systems.

Question 17

Evaluate the effectiveness of different loss functions in improving neural network performance.

Answer 17

Different loss functions, like mean squared error or cross-entropy, can be more effective depending on the type of problem (e.g., regression vs. classification) and the specific characteristics of the data.

Question 18

What is the purpose of an activation function in a neural network?

Answer 18

The activation function introduces non-linearity into the network, allowing it to learn and perform more complex tasks.

Question 19

What is the difference between supervised and unsupervised learning?

Answer 19

In supervised learning, the model is trained on labeled data. In unsupervised learning, the model learns from unlabeled data, identifying patterns on its own.

Question 20

What is a deep neural network?

Answer 20

A deep neural network is a neural network with multiple hidden layers, allowing it to model more complex relationships in data.

Question 21

What is a loss function, and why is it important?

Answer 21

A loss function measures the error between the predicted output and the actual

Question 22

How does batch size impact neural network training?

Answer 22

Batch size, the number of training samples used in one iteration, impacts training speed and stability. Larger batches provide a more stable gradient, but smaller batches offer faster convergence.

Question 23

Explain the challenge of local minima in neural network training.

Answer 23

Local minima are points where the model’s performance stops improving during training. The model gets ‘stuck’ and doesn’t reach the optimal solution.