Azure AI Fundamentals

MSFT recognizes 6 principles of AI

(1) Fairness (2) Reliability and safety (3) Privacy and security (4) Transparency (5) inclusiveness (6) Accountability

You created a classification model with four possible classes. What is the size of the confusion matrix?

4x4 A confusion matrix provides a tabulated view of predicted and actual values for each class. If we are predicting the classification of four classes, our matrix will be by 4x4 A confusion matrix for a good model will have most samples along the diagonal

Feature of Azure Machine Learning: Azure Machine Learning designer

A graphical interface enabling no-code development of machine learning solutions.

Working at a car dealership... boss asks to provide info about how many blue cars he needs for next quarter. You decide to create a AI model and choose unsupervised learning.

A numeric prediction is found using a regression model based on the historical sales data.

Supervised learning

Addressing datasets with labels or structure, data acts as a teacher and "trains" the machine, increasing in its ability to make a prediction or decision.

Unsupervised learning

Addressing datasets without any labels or structure, finding patterns and relationships by grouping data into clusters.

You want to build a personal virtual assistant. What services will you use to connect your assistant with various input channels and devices

Azure bot service

You created a CV model. You want the model to detect trained objects in photos. What info will you get about each object if you are using an object detection model?

Bounding box, class name, probability score

Feature of Azure Machine Learning: Data and compute management

Cloud-based data storage and compute resources that professional data scientists can use to run data experiment code at scale.

What are the four types of compute resources you can use in Azure Machine Learning Studio

Compute Instances, inference clusters, classification clusters, attached compute

Feature of Azure Machine Learning: Pipelines

Data scientists, software engineers, and IT operations professionals can define pipelines to orchestrate model training, deployment, and management tasks.

Challenge/ Risk: Solutions may not work for everyone

Example: A home automation assistant provides no audio output for visually impaired users

Challenge/ Risk: Bias can affect results

Example: A loan-approval model discriminates by gender due to bias in the data with which it was trained

Challenge/ Risk: Users must trust a complex system

Example: An AI-based financial tool makes investment recommendations - what are they based on?

Challenge/ Risk: Errors may cause harm

Example: An autonomous vehicle experiences a system failure and causes a collision

Challenge/ Risk: Accountability

Example: An innocent person is convicted of a crime based on evidence from facial recognition - who's responsible?

Challenge/ Risk: Data could be exposed

Example:A medical diagnostic bot is trained using sensitive patient data, which is stored insecurely

While working at a car dealership, boss asks to provide forecast info. Will new car model be successful or not. New model has a variety of engine improvements. You complied a list of data about previous successful models with their characteristics and sales numbers What should you do in the pre-processing data stage that would help predict the success of the new model?

Feature selection

When you are prepping data for the model training, you have to use your domain knowledge to select the label (or labels), features, and scale and normalize them What is the generic name for the process that includes all the steps mentioned above?

Featurization- includes feature selection, data normalization, model training, missing data handling

Image classification

Image classification involves training a machine learning model to classify images based on their contents

What is feature selection?

In short, feature selection helps us narrow down the features that are important for our label prediction and discard all features that don't play or play a min role in label prediction. This makes our trained model and prediction will be more inefficient

What components do you need to create a simple Web Chat bot?

Knowledge base, utterances, bot service

knowledge mining

Knowledge mining is the term used to describe solutions that involve extracting information from large volumes of often unstructured data to create a searchable knowledge store.

What are the 5 elements of MSFT AI

Machine Learning- The foundation o f AI systems Anomaly Detection- tools and services for identification of the unusual activities Computer Vision- tools and services for understanding and recognizing objects in images, video, faces and text Natural Language Processing- Tools and services for language understanding : text, speech, text analysis and translation Conversational AI: Tools and services for intel convo

What are the four types of entities that you can create during the authoring of the LUIS application?

Machine-Learned: Entities that are learned by your model during training from context in the sample utterances you provide. List: Entities that are defined as a hierarchy of lists and sublists. For example, a device list might include sublists for light and fan. For each list entry, you can specify synonyms, such as lamp for light. RegEx: Entities that are defined as a regular expression that describes a pattern - for example, you might define a pattern like [0-9]{3}-[0-9]{3}-[0-9]{4} for telephone numbers of the form 555-123-4567. Pattern.any: Entities that are used with patterns to define complex entities that may be hard to extract from sample utterances.

Object detection

Object detection machine learning models are trained to classify individual objects within an image, and identify their location with a bounding box.

You created a Custom vision model using the CV portal. What info do you need to provide developers to use this model?

Project ID Model Name Prediction Key Prediction End point

The application scans a document and returns the following for each page: page info, lines info, and words for each line with a confidence level What API does the application use to scan the doc?

Read: helps "read" within in predominantly doc images. asycn service desigend for the heavy text images of docs with lots of disroitions. Produces a result that includes: page info for eahc page including

Reinforcement learning

Replacing the human operator, an agent—a computer program acting on behalf of someone or something—helps determine outcome based upon a feedback loop.

Semantic segmentation

Semantic segmentation is an advanced machine learning technique in which individual pixels in the image are classified according to the object to which they belong

You are working at a hotel chain. You are planning to apply NLP to the sentiment analysis of the customer reviews. What sentiment score should you expect for the following review: The prices were ridiculously high! We could stay at the palace for half the price! The water in shower was cold

Sentiment analysis is b/w 0 and 1 0 means negative, 1 positive .5 nuetral

You build a bot framework and azure bot services. You want to extend the capabilities of your bot. What will you use to achieve goals?

Skills

Cognitive Service: Face

The Face service enables you to build face detection and facial recognition solutions.

Feature of Azure Machine Learning: Automated machine learning

This feature enables non-experts to quickly create an effective machine learning model from data.

Natural language processing: Azure Bot

This service provides a platform for conversational AI, the capability of a software "agent" to participate in a conversation. Developers can use the Bot Framework to create a bot and manage it with Azure Bot Service - integrating back-end services like Language, and connecting to channels for web chat, email, Microsoft Teams, and others.

You created an AI solution and provide info on the POSSIBILITIES and LIMITATIONS. What principle of AI?

Transparency- helps people understand how to use AI solutions, their behavior, possibilities and limitations

Natural language processing: Language

Use this service to access features for understanding and analyzing text, training language models that can understand spoken or text-based commands, and building intelligent applications.

Cognitive Service: Form Recognizer

Use this service to extract information from scanned forms and invoices.

Natural language processing: Speech

Use this service to recognize and synthesize speech, and to translate spoken languages.

Cognitive Service: Custom Vision

Use this service to train custom image classification and object detection models using your own images.

Natural language processing: Translator

Use this service to translate text between more than 60 languages.

Image analysis

You can create solutions that combine machine learning models with advanced image analysis techniques to extract information from images, including "tags" that could help catalog the image or even descriptive captions that summarize the scene shown in the image.

Cognitive Service: Computer Vision

You can use this service to analyze images and video, and extract descriptions, tags, objects, and text

LUIS

a ML based service to build NLP into apps, bots and IoT devices

Regression and classification modeling are two parts of ...

supervised ML

Clustering belongs to ___ ML

unsupervised