Artificial Intelligence (AI) and Machine Learning(ML) are one of those buzzwords that became really popular in recent years. How much hype vs. substance is behind them? Are they practical? How can you use them to solve your specific GIS problems?
Those were just a few questions that had in mind and wanted to answer.
Fortunately, I found devlopmentSEED which specialise in using Machine Learning to solve important problems and ask them to help me out.
I had privilege to talk to Drew who works there and he was eager to answer all of my questions:)
If you’re interested in AI/ML and especially how you can use it in your GIS work, this interview is a great way to start!
00:00 Short intro: what’s this interview is all about?
00:53 How Developmentseed makes Machine Learning more accessible for everyone?
02:06 What’s Machine Learning anyway and how it relates to Artificial Intelligence and GIS?
03:50 How much data do you need to make Machine Learning practical?
05:29 What we can do with Machine Learning that we can’t actually do without it? Drew tells a story about how ML helped mappers with mapping Energy infrastructure in developing countries.
09:32 Why IA makes more sense that AI right now?
10:20 What’s Developmentseed’s Skynet? (and why you don’t need to fear the evil AI)
11:47 Should GIS Professional learn Machie Learning?
13:35 What would be the very first step for GIS Professionals to start learning Machine Learning?
16:04 What’s LabelMaker and why should you try it?
18:53 When will Machine Learning be integrated into popular GIS desktop software?
20:18 Any ML use cases not based on OSM?
22:05 What’s the main chellange in “training” your own ML algorithms and how to overcome them?
25:21 Question to you dear GIS Professional: How do you think ML would help you with your GIS workflow?
You can also listen to the interview on youtube: https://youtu.be/Nw5bg2G99bM
- Label Maker: https://github.com/
- Label Maker blog post: https://developmentseed.
- Label Maker on AWS: https://developmentseed.
- Label Maker with TensorFlow: https://
developmentseed.org/blog/2018/ 02/13/tensorflow-object- detection-case/
- Mapping Pakistan, Nigeria, and Zambia Electricity Infrastructure: https://
Machine Learning resources:
- Google Machine Learning Glossary: https://developers.google.com/machine-learning/glossary/
- Flashcards: https://machinelearningflashcards.com/
- Data Science Glossary: http://www.datascienceglossary.org/
- CS231 notes from Andrej Karpathy: http://cs231n.github.io/
Question of the day: How do you think Machine Learning would help you with your GIS workflow?
Leave me a comment below, thank you!