Since it's founding, Tebex has processed over 18 million payments for games servers around the world. This is a rich data source, and one that we've significantly under-used. We know there should be ways of using this data to improve our product and to make our merchant's lives easier - how and where to start, now that's a totally different question! Join Liam as he shares Tebex's initial experiences in experimenting with machine learning to use past data to prevent fraud in the future - different algorithms we've tried, (many) failed experiments, the situation today and plans for the future.

Comments

Please login to leave a comment

Rated 5

Iwona at 19:18 on 21 Feb 2019

Really informative and interesting!

Rated 5

Christophe at 22:58 on 21 Feb 2019

Very good. Nice to hear some people are struggling with serious maths too.

The font was too small on some slides or its contrast too low, last slide.

Please keep doing live demos, they bring excitement and realism into talks.

A very interesting talk, easy to follow for someone with no experience in the subject. Well delivered with some light hearted jokes throughout.

This talk helped to put in context the core concepts of machine learning. I have seen previous talks on the subject that created as many questions as they answered. The real world example has helped cement the concepts for me.

The flow of the talk through the timeline of research, experimentation and testing showed a good approach to follow for those new to the subject, a good learning path.

Good delivery interspersed with jokes that made a heavy subject easier to follow

An interesting approach to using PHP for ML. A unique talk with a real case example.

Rated 5

Erik Bogaerts at 13:11 on 22 Feb 2019

Best introduction to machine learning I have come across. Excellent!

Rated 5

TuKlyepri at 16:04 on 22 Feb 2019

*