Please note that anyone is eligible to participate, regardless of study background and previous coding and AI experience.
<aside> 💡 For any and all questions regarding the AI Moonshot, please feel free to reach out to us at industry@tum-ai.com.
</aside>
Challenge is provided by Eyeo, the company behind two of the world’s most popular ad filtering extensions (AdBlock and AdBlock Plus)
Winners will receive a certificate of participation, letters of recommendation (for all team members), some Eyeo goodies, the chance to showcase project findings within the company, a cash prize of 1500 EUR and much more!
Online ad filtering relies on community filter lists, which contain rules for hiding HTML DOM elements responsible for rendering ads
These lists can become huge and are difficult to manually maintain over time, and are only available in certain languages
This project aims at harnessing generative machine learning to create these filter rules automatically for a given web page
You will receive at least 80 thousand webpage HTML filter rule pairs to train your models
The test dataset contains an additional 20 thousand web page HTMLs, and will be used to evaluate submissions
The evaluation is based on the average percentage of overlap between predicted rules and rules present in the gold standard datasets
You can read and download the full project description by Eyeo below.