eyeo pioneers use of machine learning in ad filtering
July 28, 2022
the teams at eyeo have successfully deployed an effective machine learning solution to counter online circumvention of ad blockers, one of the biggest threats to achieving accurate ad-filtering. While some websites try to evade ad-blocking or ad-filtering technology, eyeo’s work with machine learning models, while recognizably challenging, allows it to quickly respond to evasive measures and continue to deliver ad filtering that respects the user experience while providing value to publishers and advertisers.
Pioneering this technology gives eyeo’s ad-filtering engine a competitive advantage over other ad blockers by being able to respond to these types of disruptions in a more robust way. Machine learning solutions are able to generalize i.e. detect ads that we’ve never seen before – those that classic ad filtering wouldn’t detect. Also machine learning solutions are more difficult to circumvent.
Circumvention of ad blockers and then blocking that circumvention is a continual cat-and-mouse game, one without a roadmap to follow. However, as part of eyeo’s goal to innovate, experiment and provide a better user and privacy-friendly online experience to as many users as possible, eyeo’s set one huge goal – a moonshot – earlier last year.
Project Moonshot is its current foray into applying machine learning to our ad-filtering core to give partners cutting edge options to enhance their own products. By automating ad detection, eyeo hopse to filter the more intrusive ads while reducing human intervention of filter list creation and maintenance. eyeo can also optimize ad blocking and ad filtering on mobile platforms and speed up anti-circumvention efforts.
To reduce time and resources spent, eyeo’s team automated the pipeline to automate the collecting and the preprocessing of the data needed to train the models. This gives developers more time to work on other important breakthroughs like using machine learning to quickly and effectively solve circumvention challenges as they appear.
“We took something tried and tested, such as the model architecture and machine learning methods, and deployed them in a brand new way in Project Moonshot. What’s interesting is that the way we collect the data and train for our specific use case has never been done commercially before. By going in this direction, we are able to offer ad-filtering technology in a way that nobody else can. I feel that these are the first small steps for automating ad filtering, but it’s one giant leap for the online world,” Humera Noor Minhas, Director of Engineering at eyeo.