Building an End-to-End Hackathon AI Solution

Slide decks for NASA Space Apps Challenge Port Harcourt's hackathon (2019), one of the most formidable hackathons in the world.

This slide took participants through the basic concept of artificial intelligence and how we could use it to solve very complex problems in our hackathon.

- At slide 23 ("Launching Into Our Solution!!!"), we "launched" into our solution which was to build an image recognition system that can predict the type of flooding that might occur in a certain area based on the land topology. We used Google Cloud and pre-trained models to achieve ~70% classification accuracy on the test set.

- The slides also contain resourced on where and how to get started in AI and machine learning.

Transcript

1. Building an End-to-End Hackathon AI Solution. “Why Write 100*100 Lines of Code To Change The World When You Can Write 100*3 To Do It?” - Some Random Geek.
2. Two Questions... • Why Artificial Intelligence (AI)? • What is Artificial Intelligence (AI)?
3. Why AI? • The need for a technology that extends the human’s abilities in an intelligent manner. • The need for a technology the most complex problems for humanity. • The need for a technology that offers increased efficiency.
4. What is AI? … Ability of machines to simulate or mimic human intelligence.
5. It is powered by a technology termed “Machine Learning”...
6. What do you think when you hear the term “Machine Learning”?
7. Why Machine Learning?
8. How would you write a program to classify an email to be either spam or ham (not spam)? Quick Exercise...
9. Taken from Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 2nd Edition
10. Email Spam NotSpam
11. Taken from Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 2nd Edition
12. What is Machine Learning?
13. Using data to answer questions. ML-generated Answers. ML Model (or Algorithm Data ML
14. Machine Learning Project Workflow/Checklist.
15. 1. Frame the problem and look at the big picture. 2. Get the data. 3. Explore the data to gain insights. 4. Prepare the data to better expose the underlying data patterns to Machine Learning algorithms. 5. Explore many different models and shortlist the best ones. 6. Fine-tune your models and combine them into a great solution. (Model evaluation & testing occurs here.) 7. Present your solution. 8. Launch, monitor, and maintain your system.
16. Bonus: Difference Between Traditional Software Engineering and Software Engineering with Machine Learning (The New Age of Software).
17. Requirements Engineering Design Implementation (Write rules) Verification And Validation Maintenance ~ 100*100 Lines of Code!!! Traditional Software Engineering Phases
18. Requirements Engineering Design (Get Data; make it clean ) Implementation (ML model) Evaluation And Fine-tuning Maintenance < 100*5 Lines of Code!!! Software Engineering With ML Phases Machine Learning
19. Excited? Let's Dive Right In!!!
20. Understanding the Problem; We have been having serious problems with flooding in the island, close to the sea shores. Can you help us build a system that can detect potential areas that will be flooded before the heavy rainfall occurs? We’ve been given about 8,000 images of different sites on the island that were captured from drones, and they have been classified into categories based on their potential to suffer flood damage.
21. Understanding why flood happens and what causes them...
22. What We Will Build? A system that can classify images or footage of areas that will potentially suffer flood damage when heavy weather is incoming.
23. Launching Into Our Solution!!!
24. • Project Reference • Fast.ai • Google Cloud AI Hub • Hands-On Machine Learning with Scikit-Learn, Keras, and TensorFlow, 2nd Edition • Machine Learning Problem Framing by Google • Machine Learning by Stanford • Great Article on Learning Data Science for Free • Another Great Article Where To Learn More?
25. Final Remarks... AI technology is here to; • Extend our abilities in an intelligent manner. • Solve our most complex problems • Increase our efficiency. • Help us change the world without writing 100*100 lines of code; < 100*3 is enough.
26. THANKS FOR ATTENDING! http://bit.ly/32qbfXD https://www.linkedin.com/in/ stephenoladele/ @nerdCyberArtist Links to presentation, and profile;