Eitan Zimmeran, a data scientist at Argmax would share from his experience working on the Artlist.io recommendation system.
In addition to having the recommendation feed personalized, an additional step is required to make the feed more visual appealing.
Eitan would cover the following aspects:
- Color spaces: using RGB vs LAB to represent color distribution
- Random Walk - repeating the same colors again and again creates a "chess board" pattern, introducing randomness can help out.