Azure AI Services and APIs
What is AI?
A discussion of AI and access to AI services in the cloud
-
What is AI? Definitions and impressions from the fellowship
- What is Machine Learning?
For a classification task, say 2 class, training mathematical model input is the features, output is the class. Once the model knows what kind of features will predict the classes (based on hand-labeled). Train the model to predict the category of new data.
Unsupervised learning does not require pre-labelling, using clustering or other techniques to identify the features for classification
- How is AI different (if at all) from Machine Learning?
AI is deep learning with larger data sets, faster and more reliably.
- What are the essential ingredients to create an AI system?
- What are the bias/issues with AI
Would AI as we see it today be possible without cloud computing?
- Public AI vs Private AI: Google and Facebook have been working on this for a long time.
- 'Cloud' versus 'Internal'
What is AI or ML? Services via functions
- AI like Machine Learning in the end is a function with inputs and outputs
- Simple example: linear model from least squares ( aka regression)
Azure AI use cases
Class Discussion
Azure Cognitive Services
What are these services? Massive machine learning models developed and trained by MS for public use
- "AI Services" : https://learn.microsoft.com/en-us/azure/ai-services/
- Vision: Image detection
- Object Detection https://learn.microsoft.com/en-us/azure/ai-services/computer-vision/concept-object-detection-40
- Use case: Deer census using camera traps : https://github.com/billspat/hpcc_megadetector
Standard Practice for engagine with Azure Cognitive/AI services:
- create 'account' in your resource group
- open 'studio'
- interact visually
- learn how to use API (python)
- create code and use personal keys
OpenAI via Azure
- What is ChatGPT and how can it be used?
- API Services
- Exploring using the "OpenAI Studio"
- Azure Service https://learn.microsoft.com/en-us/azure/ai-services/openai/
- OpenAI Service: https://openai.com/product
- Azure vs OpenAI
Example of using OpenAI API
Python Program pulled from several Azure Examples:
Azure OpenAI Demo Script: https://github.com/billspat/azure_openai_demo
Customizing your own AI
- What is Transfer Learning?
- Custom training from Azure , differentiating your AI.
Acceptable Use
We have an opportunity to use these tools as part of the fellowship via the MSU Azure acccount. Please use these tools responsibly. Please see the following policies
- MSU Acceptable Use Policy: https://tech.msu.edu/about/guidelines-policies/aup/
- Azure Responsible use of AI: https://learn.microsoft.com/en-us/azure/ai-services/responsible-use-of-ai-overview