Introduction

Overview

Teaching: 15 min
Exercises: 0 min

Questions

• What will be covered in this lesson?

• What will not be covered in this lesson?

Objectives

• Set the stage for the rest of this workshop

Who we are

OCIO Data Science Lab + The Carpentries = This workshop

Carpentries Code of Conduct

Even though this is not technically a Carpentries workshop, we will be following the Carpentries Code of Conduct.

What will we be covering?

• What are AI and Machine Learning?
• What is Machine Learning good at?
• What are weaknesses of Machine Learning?
• How to use machine learning models

What will we not be covering?

• No math
• No coding

Why are we offering this seminar?

It might not seem like it, but AI can probably make your job easier right now.

As we go through the different sections today, and cover new concepts, think of how you might be able to incorporate them into your daily work.

Key Points

• This workshop will be covered by the Carpentries Code of Conduct

What are AI and Machine Learning?

Overview

Teaching: 30 min
Exercises: 15 min

Questions

• What is the difference between Artificial Intelligence and Machine Learning?

• What is the data science life cycle?

• What are some of the ethical considerations with using machine learning?

Objectives

• First learning objective. (FIXME)

What are AI and Machine Learning?

We use these 2 terms interchangeably a lot, but there are actually subtle differences.

Artificial Intelligence Definition

When a computer is able to make decisions that mimic how a human would.

AI Example: Clippy

AI Example: SPAM filter

Machine Learning Definition

A subset of AI, where decision rules are formulated from data examples.

Machine Learning is a subset of AI

(Add a new Venn Diagram here)

Clippy and SPAM filter revisited

Activity:

• What abilities would an AI-only (not using Machine Learning) thermostat have, versus a “smart” thermostat like a Nest?
• Can you think of another example of hardware or software that can be considered AI, but not necessarily based on machine learning?

What is a Machine Learning model?

Use a grading or job application rubric as example

How is a Machine Learning model built?

• First, identify the prediction to make, and how to measure its correctness.
• Examine the example data, and try to identify “features”. Domain expertise is important here.
• Either randomly subset part of the data for testing, or identify a new set of data for that.
• Train a model. Depending on the algorithm, go through an iterative process of adjusting weights to maximize your measure of correctness.
• Try it out. Do the predictions make sense? Look at near-misses. Can features be added or removed to reduce confusion?
• Loop back as needed.

Key Points

• First key point. Brief Answer to questions. (FIXME)

What is Machine Learning good at?

Overview

Teaching: 30 min
Exercises: 10 min

Questions

• What kinds of tasks do machine learning models excel at?

• What machine learning can be done on tabular data?

• What tasks are common in computer vision?

• What tasks are common in Natural Language Processing?

Objectives

• First learning objective. (FIXME)

FIXME

Key Points

• First key point. Brief Answer to questions. (FIXME)

What are weaknesses of Machine Learning?

Overview

Teaching: 30 min
Exercises: 10 min

Questions

• What types of tasks are hard for machine learning?

• How does dataset construction affect models?

• How can machine learning make biased decisions?

Objectives

• First learning objective. (FIXME)

FIXME

Key Points

• First key point. Brief Answer to questions. (FIXME)

Applying machine learning

Overview

Teaching: 15 min
Exercises: 10 min

Questions

• How can I test out machine learning on a handful of samples?

• What are ways of scaling up in machine learning?

Objectives

• First learning objective. (FIXME)

FIXME

Key Points

• First key point. Brief Answer to questions. (FIXME)

Next steps

Overview

Teaching: 15 min
Exercises: 0 min

Questions

• How can I learn how to build custom machine learning models myself?

• Where can I learn more about the ethical considerations of machine learning?

Objectives

• First learning objective. (FIXME)

FIXME

Key Points

• First key point. Brief Answer to questions. (FIXME)