Identify Common Machine Learning Techniques AI-900 Practice Questions (Page 4)

Questions 16–20

1. 16application · medium

   A company wants to build a system that recognizes objects in images, such as cars, pedestrians, and traffic signs, for an autonomous vehicle. The images are complex and contain many patterns. Which approach is most appropriate?

   • AUse a simple regression model to predict the object category.
   • BUse a transformer model with self-attention, which is specifically designed for image recognition.
   • CUse a clustering algorithm to group pixels into objects.
   • DUse a deep learning model with multiple neural network layers to learn hierarchical features from images.
   Reveal Solution Discussion Select an answer first
2. 17expert · hard

   A company wants to build a system that can read handwritten notes and convert them into digital text. The notes vary in handwriting style, lighting, and paper quality. They have a large dataset of labeled handwritten samples. Which approach should they use?

   • AUse a simple template matching system.
   • BTrain a deep neural network with multiple layers, such as a CNN or RNN, on the labeled samples.
   • CUse a regression model to predict the length of the text.
   • DUse a clustering algorithm to group similar handwriting styles.
   Reveal Solution Discussion Select an answer first
4. 18application · medium

   A marketing team wants to divide their customer base into distinct groups based on purchasing behavior, demographics, and website activity. They do not have predefined customer segments and want to discover natural groupings to tailor campaigns. Which machine learning technique should they use?

   • ARegression to predict a customer's lifetime value.
   • BDeep learning with a transformer to generate customer profiles.
   • CClustering to group customers with similar characteristics without predefined labels.
   • DClassification to assign each customer to a known segment.
   Reveal Solution Discussion Select an answer first
5. 19expert · hard

   A company wants to build a chatbot that can understand customer queries and route them to the appropriate department. The queries are in multiple languages and contain slang. Which approach should they use?

   • AUse a clustering algorithm to group similar queries.
   • BUse a Transformer-based language model fine-tuned for intent classification.
   • CUse a rule-based system with predefined keywords.
   • DUse a regression model to predict query length.
   Reveal Solution Discussion Select an answer first
6. 20expert · hard

   A medical imaging company wants to classify MRI scans into 'tumor', 'no tumor', or 'inconclusive'. They have a large dataset of labeled images, but the images are high-resolution and contain subtle patterns. They also need to provide a confidence score for each classification. Which approach should they use?

   • AUse a transformer model to process the images as sequences of pixels.
   • BUse a regression model to predict a continuous score and threshold it.
   • CTrain a deep convolutional neural network with a softmax output layer to provide class probabilities.
   • DUse a clustering algorithm to group similar images and assign labels based on cluster centroids.
   Reveal Solution Discussion Select an answer first