Norah Klintberg Sakal

Guide to your own artificial intelligence application in 3 easy steps


Audience: Beginner


Description:

How difficult is it to build an own intelligent application? When you hear artificial intelligence you might think of complex solutions as self-driving cars, autonomous drones and robots or smart assistants as Alexa or Siri - but it doesn't have to be that complex. You can build a powerful image classification model within a topic that inspires and interests you - with 3 easy steps. First, find a topic that interests you; it can be something that inspires you, excites you or a super simple day-to-day problem you're facing. Which of your skills are matching your interests and other topics that excites you? Use that skill set to the next step, which is creating your own data. My day-to-day repetitive problem was the daily makeup routine; it takes time and its quite uninspiring but when you finally decide to try something new - you literally get over 30 million results when search for makeup tutorial on Youtube and beauty being a $400 billion industry indicates that makeup play a significant part of people's lives. How do you even know what makeup style will look as good on you as it does on all the YouTubers? Therefore I built my very first machine learning algorithm which tells you which eye shape you have and suggests makeup styles matching your particular eye shape to enhances your features. You just snap a photo of your eyes in the app and the algorithm does the rest; telling you which eye shape you have and what makeup style that suits your particular eye shape. I build this app with 3 easy steps and I'm excited to show how you can build yourself an own powerful image classification model, with your own data, within a topic that inspires and interests you. It might sound difficult but I can't wait to show you how few lines of Python you need to build the whole app. This talk is for machine learning beginners - by a beginner: I built this classification model with very little experience and the emphasis of this talk is to inform that it is possible to get state of the art results on niches which lack data and that it still exists low having fruits in numerous verticals. The essence of the talk is to encourage and inspire to implement data science on niches and verticals that aligns with your own interests. What repetitive and categorizing tasks are you regularly facing? Bring your interests, passion and everyday problems and we'll explore a concrete and more widespread adoption of artificial intelligence solutions.