Diffusion models are fascinating. They take inspiration from physical phenomena and change the classical perspective of #machinelearning. Slowly destroying #structure in a #data distribution through noise and then learning a reverse #process that restores a similar structure.
We (@PiratePeel, @VBuskens, @jiamin_ou, Mahdi and me) just finished teaching a week-long summer school on #networkScience It was a lot of work but also very rewarding. All our materials are available online: https://t.co/mWNabUct2W
#TikTok challenges attract a lot of attention. As we know, some challenges are fun, while others can be dangerous. We studied dangerous and non-dangerous challenges and extracted information to characterize them.
https://t.co/AZjqF42zB3
#research#NetworkScience#DataScience
New preprint with @san_chowdhary & @fede7j from @dnds_ceu where we analyse 120 million #chess games from @lichess and compare strategies of players across different skill levels
https://t.co/T6mESqhQpV