Crowdly

Add to Chrome

Universities
learning.monash.edu
ECE2191 - Probability and AI for engineers - S2 2026

ECE2191 - Probability and AI for engineers - S2 2026

Looking for ECE2191 - Probability and AI for engineers - S2 2026 test answers and solutions? Browse our comprehensive collection of verified answers for ECE2191 - Probability and AI for engineers - S2 2026 at learning.monash.edu.

Get instant access to accurate answers and detailed explanations for your course questions. Our community-driven platform helps students succeed!

Suppose we have a model . If we evaluate on this dataset,

https://learning.monash.edu/pluginfile.php/6882264/question/questiontext/16393447/4/147960467/dataset1.csv

what is the BIC for this given model. Round your answer to 2 decimal place.

View this question

Why do we not use a mean-square-error-like error for logistic regression?

The gradient of such loss function can cause numerical instability

We cannot find find the gradient of such loss function

The mean-square error loss can lead to poor convergence when the predictions are far from the true labels due to gradients being extremely small.

✅

View this question

Why do we use one-hot vectors to represent classes rather than integers?

Integer representation implies an ordinal relationship between classes (i.e. class 0 < class 1 < class 2)

✅

One-hot vector reduces memory usage

One-hot vector is more computationally efficient.

One-hot vector is easier to interpret

View this question

Which of the following optionsd is the cross-entropy loss

View this question

What are some properties of the output of the softmax function, z

The output values are all non-negative

✅

The output values can be less than 0.

The sum of all output values equals 1

✅

The output can be interpreted as probabilities

✅

View this question

Why is grid search algorithm (V0 algorithm) infeasible for large parameter.

Potato

This is because the search space grows exponentially with the dimension of the parameter.

100%

This is because the algorithm becomes more inaccurate as the dimension of the parameter grows.

View this question

What can happens when a large learning rate is used.

The algorithm will always find the exact minimum

The algorithm will stop early

The algorithm will miss the minimum and may oscillate or diverge

✅

The algorithm will converge faster every time.

View this question

How does gradient descent improve over the grid search (V0 algorithm)?

Gradient descent allows us to explore the continuous parameter space rather than in discrete step size.

✅

Gradient descent always gurantees in finding the global minimum, while grid search may get stuck in a local minima.

This is because the derivative gives the direction of steepest ascent, allowing us to more efficiently find the optimial solution in a large and complex search space.

✅

View this question

What does it mean to train a model?

To randomly adjust the parameters of the model until it produces accurate predictions.

To find a set of parameters that minimise some loss function, which quantifies the error between the predicted and actual outcomes.

✅

To find a set of parameters that maximize some objective function, which measures the performance of the model based on the task at hand.

✅

To memorize the entire dataset in order to predict future outcomes.

View this question

Suppose we have a model . If we evaluate on this dataset,

https://learning.monash.edu/pluginfile.php/6882264/question/questiontext/16393404/4/147960467/dataset1.csv

what is the BIC for this given model. Round your answer to 2 decimal place.

View this question

Want instant access to all verified answers on learning.monash.edu?

Get Unlimited Answers To Exam Questions - Install Crowdly Extension Now!

Add to Chrome

Telegram Instagram TikTok Question Bank

Add to Chrome