Your AI algorithm can be as confident as you want it to be! The key? Let AI handle what it’s confident at and involve humans for the hardest cases. It turns out, you can dramatically gain in accuracy for the price of little indecisions! https://t.co/VFKfhkTV71
If you’re at the @InstMathStat APRM
conference in Melbourne this week, be sure to check out session IP48: "Estimation and inference under constraints".
Joseph Lawson, @Ndaoud12, and myself will be sharing some of our work this Sunday at 3:30pm.
🚨 BREAKING: Google DeepMind just revealed Gemini- ChatGPT's biggest competitor.
Gemini is the FIRST multimodal AI to outperform human experts on the MMLU, scoring over 90%.
Dr. Chao Gao, University of Chicago, receives the 2021 IMS Tweedie Award “for groundbreaking contributions to robust statistics, including establishing connections with generative adversarial networks, network analysis, and high-dimensional statistical inference.”
@ccanonne_ This looks like a rescaled variance of p(X). Is this quantity more natural than skewness to capture non uniformity ? Although skewness is more related to asymmetry...
@neu_rips@SebastienBubeck Maybe to get an alternative to this upper bound you can use a first order Taylor expansion for f between 0 and X then upper bound spectral norm by trace. Assuming that f(0)=0 you end up having a new bound of order d instead of sqrt(d) but it is indep of the condition number.
The image analysis #algorithm behind the magnificent #blackHole image is mainly based on a #GaussianMixtureModel#GMM designed for high-dimensional data, namely a mixture of probabilistic PCA! https://t.co/lekBnrJbha
"Interplay of minimax estimation and minimax support recovery under sparsity", by Mohamed Ndaoud, CREST PhD candidate in statistics, awarded Best Student Paper at The 30th International Conference on Algorithmic Learning Theory (ALT-2019), Chicago, March 22-24, 2019.