View Full Microsoft AI-901 Exam Dumps and Practice Test Dumps

 

Question 81. Which Azure AI service can analyze an image and identify objects within it?

1. Azure AI Speech
2. Azure AI Vision
3. Azure AI Translator
4. Azure AI Language

Correct Answer: 2. Azure AI Vision

Explanation:

Azure AI Vision provides capabilities for analyzing visual content and extracting useful information from images. It can identify objects, generate descriptions, detect visual features, create tags, and perform other image-analysis tasks. For example, a retail application could analyze a photograph and identify products or other objects present in the image. Azure AI Speech is designed for speech and audio processing, while Azure AI Translator handles language translation. Azure AI Language focuses on understanding and processing text. Therefore, when a scenario requires an Azure service to understand the contents or characteristics of an image, Azure AI Vision is the appropriate service.

Question 82. Which technology extracts printed or handwritten text from an image?

1. Sentiment analysis
2. Machine translation
3. Optical character recognition
4. Speech synthesis

Correct Answer: 3. Optical character recognition

Explanation:

Optical character recognition, commonly called OCR, is used to recognize text contained within images or scanned documents and convert it into machine-readable text. This allows applications to process information that originally exists only in visual form. OCR can be useful for reading signs, scanned pages, receipts, forms, and other documents. Once extracted, the text can be searched, stored, translated, or analyzed using other services. Sentiment analysis evaluates opinions in text, machine translation converts text between languages, and speech synthesis generates audio from written text. Therefore, when an AI-901 scenario asks about extracting text from an image, OCR is the correct technology.

Question 83. Which Azure AI Language capability identifies people, organizations, and locations in text?

1. Named entity recognition
2. Key phrase extraction
3. Sentiment analysis
4. Language detection

Correct Answer: 1. Named entity recognition

Explanation:

Named entity recognition, or NER, identifies important entities within unstructured text and assigns categories to them. Common examples include people, organizations, locations, dates, products, and other recognized entity types. For example, if an article mentions a company, a city, and a person’s name, NER can identify those items and classify them accordingly. Key phrase extraction instead identifies important concepts or phrases, while sentiment analysis evaluates the opinion or emotional tone of text. Language detection determines which language is being used. Therefore, when an application needs to identify and categorize named entities in written content, named entity recognition is the appropriate Azure AI Language capability.

Question 84. Which Azure AI Language feature determines whether customer feedback is positive or negative?

1. Entity recognition
2. Key phrase extraction
3. Language detection
4. Sentiment analysis

Correct Answer: 4. Sentiment analysis

Explanation:

Sentiment analysis evaluates the opinion or emotional tone expressed in written content. It can help determine whether text expresses a positive, negative, or neutral sentiment. Organizations can use this capability to analyze customer reviews, surveys, support messages, social media content, and other feedback. For example, a company could process thousands of product reviews to identify general customer sentiment without manually reading every review. Entity recognition focuses on identifying people, organizations, and other entities, while key phrase extraction identifies important concepts. Language detection identifies the language of the text. Therefore, sentiment analysis is the correct choice when the requirement involves determining the positive or negative nature of customer feedback.

Question 85. Which Azure AI Speech capability converts spoken words into text?

1. Text-to-speech
2. Speech-to-text
3. OCR
4. Image analysis

Correct Answer: 2. Speech-to-text

Explanation:

Speech-to-text converts spoken language into written text. Azure AI Speech can process speech from microphones, recordings, and other supported audio sources and generate a text transcription. This capability is useful for applications such as meeting transcription, voice-controlled applications, customer service analysis, accessibility solutions, and voice search. Text-to-speech performs the opposite process by converting written text into spoken audio. OCR extracts text from images, while image analysis examines visual information. Therefore, when a scenario requires an application to understand or transcribe what a person says, speech-to-text is the appropriate Azure AI Speech capability.

Question 86. Which Azure AI Speech capability converts written text into spoken audio?

1. Speech-to-text
2. OCR
3. Text-to-speech
4. Named entity recognition

Correct Answer: 3. Text-to-speech

Explanation:

Text-to-speech converts written text into synthesized spoken audio. Azure AI Speech can generate speech from application-provided text, making it useful for virtual assistants, accessibility applications, navigation systems, educational tools, and automated announcements. The process is essentially the reverse of speech-to-text, which converts spoken language into written text. OCR is used to extract text from visual content, while named entity recognition identifies important entities in written language. Therefore, if an application already has written information and needs to communicate that information through audio, text-to-speech is the correct Azure AI Speech capability.

Question 87. Which Azure service translates text between supported human languages?

1. Azure AI Vision
2. Azure AI Speech
3. Azure AI Translator
4. Azure Machine Learning

Correct Answer: 3. Azure AI Translator

Explanation:

Azure AI Translator provides machine translation capabilities for converting text from one supported human language into another. It can be integrated into applications that need multilingual communication, such as websites, customer-support systems, business applications, and communication platforms. For example, an application could receive a message in one language and translate it before displaying the content to another user. Azure AI Vision focuses on image analysis, Azure AI Speech handles speech-related tasks, and Azure Machine Learning provides tools for machine learning development and deployment. Therefore, when a requirement specifically involves translating written content between languages, Azure AI Translator is the appropriate Azure service.

Question 88. Which Azure AI service is designed to extract information from invoices and receipts?

1. Azure AI Search
2. Azure AI Document Intelligence
3. Azure AI Translator
4. Azure AI Speech

Correct Answer: 2. Azure AI Document Intelligence

Explanation:

Azure AI Document Intelligence is designed to analyze documents and extract structured information from them. It can be used with documents such as invoices, receipts, forms, and other supported business documents. For example, an invoice-processing application can extract fields such as vendor information, invoice number, dates, totals, and other relevant values. This reduces the need for manual data entry and allows extracted information to be processed by downstream applications. Azure AI Search is primarily concerned with searching indexed information, Translator handles language translation, and Speech processes audio. Therefore, document field extraction from invoices and receipts is an appropriate use case for Azure AI Document Intelligence.

Question 89. What is the primary purpose of Azure AI Search?

1. To train deep learning models
2. To generate speech from text
3. To identify faces in images
4. To search and retrieve information from indexed data

Correct Answer: 4. To search and retrieve information from indexed data

Explanation:

Azure AI Search provides search and information-retrieval capabilities for applications. Data can be indexed so that users or applications can efficiently search for relevant information. It is particularly useful for enterprise search solutions and generative AI applications that need to retrieve relevant content from organizational data. For example, a company could index internal documents and allow employees to search for policies, procedures, or technical information. Azure AI Search is not primarily a speech-generation or face-detection service, nor is it a general-purpose model-training platform. Therefore, searching and retrieving information from indexed data is the primary purpose described in this scenario.

Question 90. What is the purpose of grounding a generative AI model with external information?

1. To provide relevant context for generating a response
2. To increase the resolution of images
3. To convert audio into text
4. To automatically replace model training

Correct Answer: 1. To provide relevant context for generating a response

Explanation:

Grounding provides a generative AI model with relevant information from external sources so that its response can use that information as context. External information can come from organizational documents, databases, search indexes, or other approved data sources. This is particularly useful when applications need to answer questions using company-specific information that may not be available in the model’s general training knowledge. Grounding does not eliminate the need for model training and is unrelated to image resolution or speech transcription. In AI-901 scenarios, grounding is commonly associated with retrieval-based solutions where relevant information is supplied to the model before it generates an answer.

Question 91. Which Azure Machine Learning capability can automatically test different algorithms and configurations?

1. Model registry
2. Automated ML
3. Online endpoint
4. Data asset

Correct Answer: 2. Automated ML

Explanation:

Automated Machine Learning, often called Automated ML or AutoML, helps simplify model development by automatically experimenting with different algorithms, configurations, and preprocessing approaches for supported machine learning tasks. It can evaluate candidate models against selected metrics and help identify suitable approaches. This can reduce the amount of manual experimentation required by data scientists and developers. A model registry has a different role because it manages registered model artifacts and versions. An online endpoint is used to expose a trained model for inference, while a data asset provides a managed reference to data. Therefore, automatically experimenting with machine learning approaches is a key purpose of Automated ML.

Question 92. What is the purpose of compute resources in Azure Machine Learning?

1. To translate text
2. To identify objects in images
3. To provide processing resources for machine learning workloads
4. To analyze customer sentiment

Correct Answer: 3. To provide processing resources for machine learning workloads

Explanation:

Compute resources provide the processing capacity needed to perform machine learning workloads. In Azure Machine Learning, compute can be used for tasks such as model training, experimentation, data processing, and inference. Different compute options can support different workload requirements, allowing organizations to select resources based on processing needs. Translation is handled by Azure AI Translator, image analysis by Azure AI Vision, and sentiment analysis by Azure AI Language. Compute itself does not provide these specialized AI capabilities. Therefore, if an AI-901 question describes the need for processing power to run machine learning jobs, Azure Machine Learning compute resources are the relevant concept.

Question 93. What is the purpose of a data asset in Azure Machine Learning?

1. To generate speech
2. To manage a reference to data used in machine learning workflows
3. To translate documents
4. To detect harmful images

Correct Answer: 2. To manage a reference to data used in machine learning workflows

Explanation:

An Azure Machine Learning data asset provides a managed way to reference and organize data used by machine learning workflows. Data assets can make datasets easier to identify, reuse, and manage across experiments, jobs, and pipelines. This helps teams maintain a consistent approach to the data used during machine learning activities. A data asset is not the same as a speech-generation capability, translation service, or content-moderation service. Those tasks are handled by other Azure AI services. In an AI-901 scenario involving the organization and reuse of data within Azure Machine Learning, the concept of a data asset is especially relevant.

Question 94. What is the main purpose of the Azure Machine Learning model registry?

1. To analyze images
2. To translate text
3. To manage registered machine learning models and their versions
4. To create speech from text

Correct Answer: 3. To manage registered machine learning models and their versions

Explanation:

The Azure Machine Learning model registry provides a way to register, organize, and manage machine learning models and their versions. Model versioning is important because machine learning teams may develop multiple versions of a model and need to identify which version was tested, approved, or deployed. Maintaining registered models supports better model lifecycle management and helps teams keep track of machine learning artifacts. The model registry does not perform image analysis, language translation, or speech synthesis. Those capabilities belong to specialized Azure AI services. Therefore, when a scenario asks about organizing and managing machine learning model versions, the Azure Machine Learning model registry is the appropriate feature.

Question 95. Which dataset should generally be reserved for final evaluation of a machine learning model?

1. Training dataset
2. Validation dataset
3. Test dataset
4. Feature dataset

Correct Answer: 3. Test dataset

Explanation:

The test dataset is generally reserved for evaluating a machine learning model after development and tuning have been completed. It should contain examples that the model has not used during training and that have not been repeatedly used to make development decisions. This helps provide a more reliable estimate of how the model may perform on unseen data. Training data is used to learn model parameters, while validation data can help compare configurations and tune hyperparameters during development. Keeping test data separate reduces the risk of producing an overly optimistic evaluation. Therefore, the test dataset is normally the appropriate dataset for final model evaluation.

Question 96. What does a confusion matrix show for a classification model?

1. The relationship between predicted and actual classes
2. The amount of cloud storage used
3. The speed of translation
4. The quality of synthesized audio

Correct Answer: 1. The relationship between predicted and actual classes

Explanation:

A confusion matrix provides a detailed summary of classification predictions by comparing the classes predicted by a model with the actual classes. For binary classification, it commonly includes true positives, true negatives, false positives, and false negatives. These values can be used to calculate important evaluation metrics such as accuracy, precision, recall, and F1 score. A confusion matrix therefore provides more detail than simply reporting an overall accuracy value. It does not measure cloud storage, translation speed, or audio quality. For AI-901, understanding confusion matrices is useful when evaluating how a classification model handles correct predictions and different types of errors.

Question 97. Which metric measures the percentage of predicted positive results that are actually positive?

1. Recall
2. Accuracy
3. Precision
4. MAE

Correct Answer: 3. Precision

Explanation:

Precision measures the proportion of predicted positive cases that are actually positive. It can be calculated by dividing true positives by the total number of predicted positives, which includes both true positives and false positives. Precision is particularly important when false-positive results can create significant costs or consequences. Recall has a different focus because it measures the proportion of actual positive cases that the model successfully identifies. Accuracy measures the proportion of all predictions that are correct, while mean absolute error is generally used for regression. Therefore, when an AI-901 scenario asks how reliable a model’s positive predictions are, precision is the appropriate metric.

Question 98. Which metric measures the average absolute difference between predicted and actual numerical values?

1. Mean absolute error
2. Precision
3. Recall
4. Accuracy

Correct Answer: 1. Mean absolute error

Explanation:

Mean absolute error, or MAE, is commonly used to evaluate regression models. It calculates the average absolute difference between predicted values and their actual values. Because the differences are converted to absolute values, positive and negative errors do not cancel each other out. MAE is expressed in the same general units as the predicted target, which can make it relatively straightforward to interpret. Precision, recall, and accuracy are primarily classification metrics. Therefore, if a model predicts numerical values and the requirement is to measure the average magnitude of its prediction errors, mean absolute error is an appropriate evaluation metric.

Question 99. Which responsible AI principle focuses on designing systems that can serve people with different abilities and circumstances?

1. Accountability
2. Inclusiveness
3. Transparency
4. Reliability

Correct Answer: 2. Inclusiveness

Explanation:

Inclusiveness is a responsible AI principle that focuses on ensuring AI systems can be used effectively by people with different abilities, backgrounds, and circumstances. Developers should consider the diversity of intended users and avoid unnecessary barriers that could prevent certain people from benefiting from an AI system. Accessibility considerations are an important example of inclusiveness. Accountability is concerned with responsibility for AI systems and their outcomes, transparency focuses on communicating how systems work and their limitations, and reliability concerns dependable operation. Therefore, when an AI-901 question focuses on making AI systems accessible and useful to a broad range of people, inclusiveness is the relevant responsible AI principle.

Question 100. Which practice can help identify and reduce bias in a machine learning solution?

1. Evaluate model performance across relevant groups
2. Ignore differences in training data
3. Remove validation and test data
4. Always select the largest model

Correct Answer: 1. Evaluate model performance across relevant groups

Explanation:

Evaluating model performance across relevant groups can help identify whether an AI system produces significantly different outcomes for different populations. Developers can examine training data, labels, features, and evaluation results to identify potential sources of unfairness or underrepresentation. If disparities are discovered, teams can investigate the causes and consider improvements to the data, model, features, or evaluation process. Simply selecting a larger model does not guarantee fairness, and removing validation or test data makes meaningful evaluation more difficult. Therefore, examining performance across relevant groups is an important practice for identifying potential bias and supporting the development of more responsible machine learning systems.