Max Wasserman

@sp_monte_carlo This aligns with the observation that pointwise evaluation of the posterior is of limited use.

@sp_monte_carlo Answering queries about unobserved variables requires integration. Integration of a general probability measure is harder than local approximation (differentiation) because it involves the entire space, which in general has uncountable support.

In most introductory augments, posterior inference is deemed intractable in non-toy settings due to the integral involved in the marginal likelihood p(x) = \int p(x,z) dz. But even with p(x) in hand, we can only evaluate the posterior p(z|x) = p(x, z) / p(x) pointwise.

Graph Structure Learning with Interpretable Bayesian Neural Networks Max Wasserman, Gonzalo Mateos. Action editor: Fabio Stella. openreview.net/forum?id=2noXK… #graphs #graph #deep

@betanalpha @sp_monte_carlo @rlmcelreath @roydanroy @sirbayes @DavidDuvenaud @tcarpenter216 @bayesian_stats @SolomonKurz These links make me want to go into a cave - armed with a pen, paper, and Papa Rudin - and only emerge when I can extracted every ounce of wisdom from your writings.

@sp_monte_carlo I have a feeling I'll be chewing on this response for a while. Thanks!

Bayesian folks, what am I missing? @sp_monte_carlo @rlmcelreath @roydanroy @sirbayes @DavidDuvenaud @tcarpenter216 @betanalpha @bayesian_stats @SolomonKurz

From what I understand, successful inference typically means “we can now approximate expectations of functions of these unobserved variables z”. This requires posterior samples. So p(x) is not really the obstacle to successful inference here.

@StatMLPapers I'll be presenting this work in NYC at the Annual ML Symposium on Oct. 18th. I'll be around for a few days before/afterward—feel free to reach out if you're in town and want to chat! events.nyas.org/event/ml2024/ @NYASciences @gonchimateos @npparikh @dshieble @sirbayes @phanein

@StatMLPapers Proud of this work. The Kumaraswamy has a key role to play in scalable latent variable modeling, but use has been limited by unstable implementations. We stabilize it and introduce the Variational Bandit Encoder—a simple, scalable approach to contextual multi-armed bandits.

Stabilizing the Kumaraswamy Distribution ift.tt/SqtXLv8

@npparikh I now think proper ordering (e.g. baby Rudin before Munkres), periodization (push one area hard at a time, maintain knowledge in others), and a supportive social context (being around people who value such endeavors) are key. Properly executing all 3 is hard.

Pick a set of topics and stick to them, don’t try to learn 40 topics at the same time, and just accept that it will take maybe 5 years. I went to a PhD partly because you just need a long period of uninterrupted time to learn certain things.

This randomly popped into my feed again in response to some post I made last year about books I recommended. It was a lot of books. But it really is possible to work through a ton of books in a relatively short amount of time and this is a thing universities are for.

@sp_monte_carlo Hot off the presses. arxiv.org/abs/2406.14786

@sp_monte_carlo Check out "Model-Based Deep Learning" nowpublishers.com/article/Detail… If pdf is needed, dm me. I have a cool paper coming out on this topic. I link unrolling params to properties of the data, allowing for (interpretable) Bayesian modeling over them. You’ll enjoy it. ~2 weeks to arxiv.

@Maxematician @sp_monte_carlo For what it’s worth, this is how’s its introduced in Baby Rudin.

@sp_monte_carlo One could argue that the term Banach space should instead refer to more general complete metric spaces.

A professor told me they don’t teach the Banach fixed point theorem at MIT for some reason. Literally, this is obviously false. But the root of what he was trying to say was that more people should understand the fixed point theorem is THE essential property of Banach spaces.

@prof_g I never took your class. You met with me when I reached out to ask silly questions on learning math when I didn’t know up from down. Your kindness, approachability & eccentricity acted as motivation to give the ‘hard’ stuff a shot. And I’m still going!

wow! i just got an email from a student who thanked me... for the way i gave them a D in calculus... 27 years ago... said that it's been on her mind to thank me for being gentle & kind about it... she wanted to tell me that she retook the course, got an A, became a scientist, then a lawyer, & now is a teacher! i'm so proud of her. i remember she was so stressed out about that grade. be kind: everything is eternal.

@alfcnz @RPstrength SS will take you on a 2 year tangent of ‘hyppe drives’, ‘sets of fahves’, and drinking gallons of milk. youtu.be/yCsvKNdxH4s?si…

@alfcnz @RPstrength SS stuff will focus on athletics, not changes in body composition (the, err, pear shape you’ve referred to). SS isn’t bad, but doesn’t compare to RP for your goals. RP does a good job making clear what does & doesn’t matters in a non-ideological, minimally overwhelming way.

Do you lift weights? 🏋🏼 Are you based in NYC? 🗽 Are you willing to welcome a newbie? 👶🏼 What book(s) would you recommend for a beginner body builder? I currently look like a pear 🍐 and I’d rather change shape 😅 but I have zero clues where to start. 😭 Thanks, Alf ❤️