thoughtt ...thinking aloud

I made a game to better understand CNNs

One of my hobbies is to play with pre-trained models to see how good they perform. Almost always it was clear to me that pre-trained models, like DeepLab or MaskRCNN, are not particularly good at detecting some abstract representations of an object. They are to some extends good at detecting (full) silhouettes, but they perform poorly when they ... Read more

A Quest for fixing Recommendation Systems

Do you ever wonder where youtube will take you if you watch 1000 videos in a row, or what happens if you press the Next button for a large number of times? Let say you start watching the trailer of Spider-Man: Far From Home, could we say anything particular about the 1000th video after this one? Is it going to be a Movie trailer or Music Video, ... Read more