Identify Common Machine Learning Techniques AI-900 Practice Questions (Page 7)

Questions 31–34

1. 31application · medium

   A company wants to build a system that can transcribe audio recordings into text and then translate the text into multiple languages. The system must handle different accents and languages. Which approach is most appropriate?

   • AUse deep learning models, such as transformers, for speech recognition and translation.
   • BUse regression to predict the length of the audio.
   • CUse clustering to group similar audio segments.
   • DUse a simple rule-based system with predefined patterns.
   Reveal Solution Discussion Select an answer first
2. 32application · medium

   An email service provider wants to automatically sort incoming emails into 'Primary', 'Social', and 'Promotions' folders. The system must also estimate the likelihood that an email is spam. Which machine learning techniques should be used?

   • AA clustering model to group emails into folders, and a classification model to detect spam.
   • BA classification model to assign each email to a folder, and a regression model to output a spam probability.
   • CA transformer-based model to assign folders and a separate transformer to detect spam, both using self-attention.
   • DA regression model to assign each email to a folder, and a clustering model to estimate spam likelihood.
   Reveal Solution Discussion Select an answer first
4. 33application · medium

   A cybersecurity team wants to detect unusual network traffic that might indicate a security breach. They do not have examples of attacks, only normal traffic patterns. Which technique should they use?

   • ARegression to predict the number of packets per second.
   • BClassification to label traffic as 'attack' or 'normal'.
   • CA deep neural network to generate a text description of traffic.
   • DClustering to group normal traffic and flag points that do not fit any group.
   Reveal Solution Discussion Select an answer first
5. 34foundation · easy

   Which characteristic is a defining feature of deep learning?

   • ARequires all features to be manually engineered
   • BUses neural networks with multiple layers to learn complex patterns
   • COnly works with tabular data
   • DUses only a single layer of neurons
   Reveal Solution Discussion Select an answer first