Why Machine Learning projects are failing and how to prevent them from failing

Daniel Sârbe - Development Manager, LanguageCloud @ SDL

Europa room, 3rd floor

13nd November, 15:10-15:40

Machine learning algorithms work really well in situations where exists lots of ground truth data, but if you don’t have enough cases for machine learning model to train on, it is unlikely that it will be able to generalize the features needed to be successful. Except good data there are various other factors that impact the success of an machine learning project and you should evaluate really well if your organization fulfills certain requirements before you jump on a machine learning endeavor.

Usually the machine learning talks are about interesting use-cases, trendy algorithm, how you can apply ML to create a competitive advantage and rarely about what can go wrong when going on the machine learning path. In this talk, I will try try to fill that gap by discussing some sounding machine learning failures, what do they have in common, and some hints how to prevent them from failing.

Daniel Sârbe


Passioned about people and technology, interested in anything that is related to Scalability, BigData, Machine Learning, Deep Learning. Experienced Software Development Manager with a demonstrated history on building teams and scalable SaaS/Cloud products. Co-founder of BigData/DataScience Meetup Cluj and frequent speaker at technical conferences, both international ones like MesosCon Europe, T3EE and local ones like: DevTalks, IT Days and BigData/DataScience meetups. Skilled in Scalability, BigData, Machine Learning, Java, Cloud, Rest API, Software as a Service (SaaS), Continuous Delivery and Scrum.