#AcademicRunPlaylist - 6/19/24

These guys in the tide pool had the right idea, but while they were probably cooler in the water at least I was able to beat the heat with some cool talks for my #AcademicRunPlaylist (as if you needed more evidence that I'm a nerd)!

First was an amazing talk by Bill Freeman in getting computers to see faces in things at the Simons Institute for the Theory of Computing. The phenomenon of face pareidolia - seeing face-like structure around otherwise random stimuli - is a particular feature of human visual perception that computers aren't good at replicating. Freemen presents a model and dataset that aim to bridge this gap and understand what image conditions are most likely to induce pareidolia, with impressive results. Highly recommend https://www.youtube.com/watch?v=ZfQivjm8OaI

Next was an intriguing talk by Katya Scheinberg on generally defining and comparing stochastic optimization methods at DIMACS https://www.youtube.com/watch?v=_aFZlwITDtM

Next was an incredible talk by Alison Gopnik on empowerment gain as causal learning at the Simons Institute. Gopnik unites her groundbreaking work with children, research with non-human mammals, and algorithms to show why optimizing for empowerment, rather than utility or novelty, is the best approach for understanding and building causal learning agents. Highly recommend https://www.youtube.com/watch?v=VHc6f5dsvrE

Next was an important talk by Samuel Segun on constructing an afro-ethical framework for AI systems at the Institute for Science and Ethics (IWE) https://www.youtube.com/watch?v=9lLmRdcfXeQ

Next was an excellent talk by Amanda Seed on differences between human and non-human primate intelligence at the Simons Institute https://www.youtube.com/watch?v=fIHOTiLi_Rw

Next was an extremely informative symposium on the global state of computational antitrust within competition agencies at the OECD and Stanford CodeX with friends of the playlist Teodora Groza and Thibault Schrepel, Bill Kovacic, and Frederic Jenny https://www.youtube.com/watch?v=isbqt30ycy4

Next was a great talk by Edward Adelson on the future of perceptual cues models in robotics at the Simons Institute. Come for the insight on different perception problems, stay for the biting quotes on ML, such as this gem: "Whenever you do a machine learning experiment of this sort, the answer is always 87%. It always worked better than you feared and worse than you hoped. So that's what we got." https://www.youtube.com/watch?v=4HTwpLtQsiM

Next was a fantastic talk by Blase Ur on privacy in a data-driven world at Stanford University. Ur digs deep into how people perceive privacy, efforts to more transparently surface data about online tracking, and where the privacy community should focus their efforts to drive meaningful change. Highly recommend https://www.youtube.com/watch?v=3F4YI9CpYeQ

Last was a fascinating talk by Tobias Gerstenberg on counterfactual simulation in human cognition at the Simons Institute https://www.youtube.com/watch?v=NiByLo2qbXk