MACHINE LEARNING COURSE

Student workbook

THIS WORKBOOK BELONGS TO:

Welcome

In this project, you and your team will work together to design and build a machine learning model that solves a problem you care about.

The process

You will work your way through a range of activities, split across 5 modules.

1. Crash

course

3. Scoping

2. Idea

generation

4. Product development

5. Pitch

COURSE ACTIVITIES

MACHINE LEARNING IN A DAY

Activity AFacial recognition

Activity BDriverless cars

Activity CSelf-driving cars

Activity DTesting and improving your model

Activity E Spot possible problems

Activity F Filter through problems

Activity G Draft Mini elevator pitch

Activity HSourcing your data

Activity I Build your chatbot

Activity J Test your chatbot

Activity K Prepare your pitch

Optional Activity 1 - Recommendation systems

Optional Activity 2 - Natural language processing

ACTIVITY A

FACIAL RECOGNITION

COURSE

MACHINE LEARNING IN A DAY

What are the issues surrounding facial recognition technology?

What are the advantages and disadvantages of using facial recognition technology?

Are the police right to be trialling facial recognition technology to track down criminals? Are people right to be concerned?

Should schools in this country use facial recognition technology in the classroom? Explain your answer.

NAME

ACTIVITY B

DRIVERLESS CARS - DECISION MAKING

Which of these factors are most important to your team? Discuss with other students and rank these factors in order of importance. Are there any other factors that should be built into the decision making process?

COURSE

MACHINE LEARNING IN A DAY

Saving more lives

Is it important to minimise the number of people or animals killed?

Avoiding intervention

Should you avoid changing the direction of the car?

Protecting passengers

Is it more important to protect passengers than pedestrians?

Gender preference

Is it more important to save males or females?

Upholding the law

Should preference be given to those obeying the law e.g. those crossing when the green light is showing?

Species preference

Is it more important to save human lives than animals?

Age preference

Is it more important to save young people or old people?

Fitness preference

Is it more important to save people who are physically fit?

Social value preference

Is it more important to save a doctor than a criminal?

NAME

ACTIVITY C

SELF-DRIVING CARS

COURSE

MACHINE LEARNING IN A DAY

What are the issues surrounding self-driving cars?

What do you think are the most important factors for the self-driving cars decision making algorithms?

Is it morally or ethically right to pre-program these decisions? Explain you answer.

What would happen if these decisions were not pre-programmed?

NAME

ACTIVITY D

TESTING AND IMPROVING YOUR MODEL

COURSE

MACHINE LEARNING IN A DAY

Try new instructions and note down the result. Did the model correctly carry out your instructions? Why do you think your model did or didn’t understand?

Test instruction

Outcome

Correct Y/N?

NAME

ACTIVITY E

SPOT POSSIBLE PROBLEMS

Use this sheet to jot down your ideas before discussing them with the rest of your team.

COURSE

MACHINE LEARNING IN A DAY

Things that I would like to be different in my local community or my life

Someone or something I worry about

Think through your average day – what frustrations or issues do you encounter?

Things that I wish more people knew about or understood better

NAME

ACTIVITY SHEET F

FILTER THROUGH PROBLEMS

COURSE

MACHINE LEARNING IN A DAY

Use this worksheet to help you think through potential problems in more detail.

Problem

Data required (and availability)?

Ethical considerations?

NAME

ACTIVITY G

DRAFT MINI ELEVATOR PITCH

Turn problems into product ideas with a mini ‘elevator pitch’. As an example. “Our team is called ZOE. We’re creating a chatbot aimed at students to help them to revise for their physics exams in an interesting and engaging way.

COURSE

MACHINE LEARNING IN A DAY

Pitch idea

We’re creating a machine learning model aimed at...

to help them to...

by providing them with...

Our team is called...

(the target user)

(the problem / challenge)

(the possible solution)

NAME

ACTIVITY H

SOURCING YOUR DATA

COURSE

MACHINE LEARNING IN A DAY

Use this worksheet to make notes on the data your chatbot will require

What data does your chatbot require?

What labels will you use?

What preparation will your data require?

Where will you obtain the data from?

NAME

ACTIVITY SHEET I

BUILD YOUR CHATBOT

COURSE

MACHINE LEARNING IN A DAY

Use this page to sketch out your design ideas or make notes on what your chatbot needs to do.

User interface

Programming notes

NAME

ACTIVITY J

TEST YOUR CHATBOT

COURSE

MACHINE LEARNING IN A DAY

It is very unlikely that your chatbotl will be 100% accurate on its first run. Use this worksheet to record the steps you took to improve your model.

How accurate was your model and how did you measure this?

What do you think caused the inaccurate results?

How effective were these changes?

What changes did you make to your model to make it more accurate?

NAME

OPTIONAL ACTIVITY 1

RECOMMENDATION SYSTEMS

COURSE

MACHINE LEARNING IN A DAY

How well do you think Netflix or other recommendation systems know you? Why is this?

How does Netflix produce recommendations?

Can you think of any drawbacks of recommendation systems?

Can you think of any other machine learning recommendation systems?

NAME

OPTIONAL ACTIVITY 2

NATURAL LANGUAGE PROCESSING

COURSE

MACHINE LEARNING IN A DAY

What could be done to make the chatbot seem more human?

Write down your three questions for the chatbot

Did the chatbot say anything unusual or silly?

Would this chatbot pass the Turing test? Explain your answer.

NAME

OPTIONAL ACTIVITY 3

FLOWCHARTS

COURSE

MACHINE LEARNING IN A DAY

Use this page to sketch out your design ideas or make notes on what your chatbot needs to do.

NAME

OPTIONAL ACTIVITY 4

PREPARE YOUR PITCH

COURSE

MACHINE LEARNING IN A DAY

Use this worksheet to plan your content and divide up the presentation to allocate sections to each member of the team.

Our problem

Our machine learning model

Our prototype

Target users

Conclusion

Team member

Notes:

Model name and slogan

Team member

Notes:

Team member

Notes:

Team member

Notes:

Team member

Notes:

Team member

Notes:

NAME

Ace, course complete!

Making great machine learning models requires hard work and constant improvement. Machine learning development is a journey. Where will your journey end?

Using these materials

Apps for Good materials are licensed under the Creative Commons Attribution-NonCommercial-ShareAlike CC-BY-NC-SA License

You are free to:

  • Share — Copy and redistribute the material in any medium or format
  • Adapt — Remix, transform, and build upon the material

Under these terms:

  • Attribution – You must attribute the material to Apps for Good
  • Non Commercial – You may not use the material for commercial purposes
  • Share Alike – If you remix, transform, or build upon the material, you must distribute your contributions under the same license as the original
AfG_Machine Learning in a Day_Student Workbook_1920_v1 - Google Slides