CENG440 Introduction to Machine Learning for Embedded Systems

In this exercise, the output depends on datasets fed to

• a machine learning model

• a set of rules

• an analytical model

• all of them

What is the primary goal of supervised learning?

• To predict future outcomes without any labeled data.

• To cluster similar data points into groups.

• To reduce the dimensionality of input features.

• To learn a mapping from input features to output labels using labeled data.

Name two common algorithms used in supervised learning.

• Neural Networks

• K-Means Clustering

• Principal Component Analysis

• Decision Trees, Support Vector Machines (SVM)

What distinguishes unsupervised learning from supervised learning?

• Unsupervised learning is only applicable to classification tasks.

• Unsupervised learning does not use labeled data, while supervised learning does.

• Both unsupervised and supervised learning use labeled data.

• Unsupervised learning requires labeled data, while supervised learning does not.

Give an example of an unsupervised learning algorithm.

• Decision tree

• Linear regression

• Support vector machine

• K-means clustering

What are some common constraints faced by embedded systems?

• Common constraints faced by embedded systems include limited processing power, restricted memory, real-time requirements, energy consumption limitations, and hardware dependencies.

• Flexible hardware dependencies

• Unlimited processing power

• No energy consumption

Why is data preprocessing important in machine learning?

• Data preprocessing only increases computation time.

• Data preprocessing is important because it enhances data quality and prepares it for effective analysis.

• Data preprocessing is unnecessary for model training.

• Data preprocessing is only relevant for deep learning models.

List two methods of model evaluation metrics.

• Log Loss, ROC Curve

• Accuracy, F1 Score

• Mean Squared Error, R-squared

• Precision, Recall

What is the main difference between classification and regression tasks in supervised learning?

• Classification predicts continuous values, while regression predicts discrete labels.

• Classification predicts discrete labels, while regression predicts continuous values.

• Both classification and regression predict continuous values.

• Both classification and regression predict discrete labels.