But on the other hand, i guess i wouldn't say that de blerning is off the table. So there was the great rise of de blerning on starting somehene in the middle of the last decade. And then was oer merging into into recommender systems fieldss, and i'd adjust remember its one of my first experiences, actually, the workshop on deep learning for recommender systems at rexus in como. So i agree with you that randomly throwing deep learning to solve a matrix factorization problem at some data is a bit too easy.

Yes, i need to cor in here and say that using deep learning to predict ratings is not the solution. And then aa about that. Ye. So there is a lot of use of deep learning in the field of recommender systems, for example, creating usa in beddings, item in bedding, vitial in beddings and so on, which you can then use in ta sense that you can do tose and beddings and then feed them into nornetworks and so on. So yet, i'm not saying that you shouldn't use deep learning. I think that there is a lot of tools that you should try out first and be sure that that's working, and then compele, and then you can try out deep learning and then see if if that actually improves anything in your system. This

was the main point. So or deep learning was wasthowing major improvements in computer vision, in natural language processing. And i guess it's having a major impact in cases where we really take into account the additional attributes that are mostly unstructured and coming with items. I remember that block post from two thousand 13 from sandadileman, who was writing about how they applied c an ns, or convolutional neal networks, to mel spectrocams, kind of visual presentations of music. And then were coming up with new dawnes, but also with visual finger prints of music, which first sounds totally awkward, because music is audio, and now you are coming up with a visual finger print. How does it work? But i guess that they really contributed a bit more to the podifire commender system there. But yer ifea may be just too narrow ly focusing on only rating prediction given a one hot ancoted user or item, then this is maybe not the right direction. And there matric factorization is more efficient, powerful, and getting you to the same resultyes, yes. Orso. Some part thatw we haven't covered so far by by going deeper into into the sids and downsights of of rex's research, was actually one of the fundamental problems, i guess. But, but we are also also dealing with in our work, which is that clibto filtering has a major problem, which is the colt start, how do you deal with it? Or if you don't want to talk about nasty colt start, then maybe let us know a bit more about the gray sheep that you are working with. A lot. Pick a topic.

Well, so one thing that i love about cold start is that every so often a new article comes out about, now we solve the cold start problem. And so basily, the cold start problem means that you don't have any data. And the way that each and every paper solves the coal star problem is by saying, ye, but we don't have anyn anything of that data, but we have some other data that we will then use to solve the coal star problem. But then actually we, if you want to go down to the definition, then you don't have a coal star problem any longer, because then you know something about eusa. So basilythe that you'd solve a coal star problem is that not that way. But some of the choicest that you can do, because i don't know if you can solve it, is you, you know a lot about on boarding. For example, you can do some non personalized er recommendations and then hope that people will catch up on it. Or you can try and do recommedatonce the other way, because in crabs filtering, need to know enough about the user such that you can find similar users out there. If you don't know anything about the user, then you can't actually do any recommendations, and then you can't do any personalization. But in the word, personalization means also you should know something about the person, because otherwise you can't actually personalize anything. So if you have something you know about this use that is different, then you can use that and compare him to other uses and then recommend what they like. Olhwise, its just a matter of doing some mix of popularity and items that are controversial, in the sense that you learn a lot from knowing whether they like this one or not.