Introduction

Question 1

What is machine learning?

Answer 1

machine learning is the way we “teach” computer model to make prediction and draw conclusions from data
- Foundation for AI solutions
ex. Yield - uses sensors, data and ML to help farmers make decisions related to weather, soil and plant conditions

MS ex. Azure Machine Learning

Question 2

Computer Vision?

Answer 2

ability for AI to interpret world visually through cameras, video, images
ex. Seeing AI app uses computer vision to provide blind/low vision community with audio description of the visual world

MS ex. Azure AI Vision

Question 3

Natural Language Processing

Answer 3

ability for AI to interpret written or spoken language and respond in kind

MS ex. Azure AI Language, Azure AI Speech

Question 4

Document Intelligence

Answer 4

ability of AI to manage, process and use high volumes of data found in forms and docs

MS ex. Azure AI Document Intelligence

Question 5

Knowledge Mining

Answer 5

capability of AI to extract info from large volumes of often unstructured data to create knowledge store

MS ex. Azure AI search

Question 6

Generative Ai

Answer 6

ability of AI to create original content in variety of formats including natural lang, image, code and more.

MS ex. Azure OpenAI Service - deploy, customize and host gen AI models

Question 7

How do machines learn?

Answer 7

They use data to learn.
Data scientists use data to train ML models to make predictions and inferences based on relations they find in data

Question 8

Give example of ML with wildflowers

Answer 8

1. data scientists will collect data on wildflower samples
2. they will label the samples with the correct species
3. Labeled data is processed using an algorithm that finds relation between features of sample and labeled species
4. results of algorithm are encapsulated in a model
5. when new samples are found the model can identify correct species label

Question 9

Image classification

Answer 9

involves training ML model to classify image based on content. ex. traffic solution might want to classify different types of vehicles like taxi, bus, cycles and so on

Question 10

Object detection

Answer 10

ml models are trained to classify several objects within an image (takes image classification to next level) and identify their correct location with a bounding box

Question 11

Semantic segmentation

Answer 11

advanced ML model that identifies individual pixels in the image are classified based on what object it is.

Question 12

Image Analysis

Answer 12

can combine ML models with advanced analysis to extract information from image that help catalog image or create some caption of what is happening in image

Question 13

Face detection, analysis and recognition

Answer 13

locates human faces in an image. Can be combined with classification and facial geometry analysis techniques to recognize individuals based on facial features

Question 14

Optical Character Recognition (OCR)

Answer 14

technique for detecting and reading text in images. i.e like stop on stop sign or name of store

Question 15

Name some challenges and risks with Ai

Answer 15

1. Bias can affect results - loan approval model discriminates based on color because of model it was trained on
2. Errors can cause harm - autonomous vehicle crashes due to a sensor failure
3. Data could be exposed - medical robot trained using sensitive patient data is stored insecurely
4. solutions not for everyone - home automation assistant has no audio output for blind
5. Trust in complex system - AI finance tool
6. Who is liable for AI decisions - innocent person convicted of crime based on facial recognition

Question 16

The focus areas to solve the challenges and risks

Answer 16

1. Fairness
2. Reliability and safety
3. Privacy and security
4. Inclusiveness
5. Transparency
6. Accountability