Assume we are building an artificial neural network for multi-class classification (there are more than 3 possible classes). Which of the following activation functions should we use in the output layer?

For this question you can assume the default case, where the model selects a class if the probability is higher than 50% and all models have the same vote in the ensembles. Assume we train 3 models for binary classification. The models return the following probabilities for the first instance/observation: Assume we use an ensemble model with Soft Voting. Which class will the ensemble model assign to this instance?

What is the purpose of dropout in neural networks?

Which of the following algorithms are supervised?

After running PCA on a dataset, you find that: Principal Component 1 (PC1) explains 40% of the variance PC2 explains 30% PC3 explains 15% PC4 explains 10% PC5 explains 5% If your goal is to keep components that explain at least 90% of the total variance, how many principal components should you retain?

In soft voting, ensemble predictions are made by:

Why are the new features created by PCA often harder to interpret?

Assume we are building a regression model using a neural network to predict house prices. Which of the following loss functions is most appropriate?

Which of the following is the most general term that includes machine learning, deep learning, and large language models?

Assume we run Lasso with a penalty weight of 0.01, and the model includes nearly all variables. If we want the model to become more selective and include fewer variables, what should we do with the penalty weight?