In ML, I typically see that the more closely a model's behaviour matches expected utility and Bayes, the better it is. It's hard to pin down the dividing line where one should, and shouldn't listen to these frameworks. I'm looking forward to the insights in the book!.

I totally agree that there are a very small number of principles that underpin all of modern-day ML. Expected utility (and Bayes as a closely related consequence) being a Big One. Both lead to strange conclusions if taken to the extreme!.

I'm excited by the prospect of this book, and worth reading if @beenwrekt's blog posts are anything to go by. "The core intellectual idea of the latest AI bubble is taking the initial ideas from the 1940s and riding them to their logical conclusion.".

In other work, we investigate metalearning as a way to implement these ideas. The advantage being that a generative model can directly learn the conditional distribution of interest, without a bottleneck of approximate inference!. For more on that, see 3/3.

This does lead to the question, what models should we use, and how should we do inference?. We use a VAE with Gaussian Process mappings (GPLVM), but the idea applies equally to Bayesian NNs, if inference can be made to work!. 2/3.

More in our investigation of using Bayesian Model Selection for Causal Discovery: Multivariate Graphs. As previously, the message is: Causal discovery requires assumptions, and Bayes enables soft, realistic assumptions. Good Bayesian inference then leads to good performance. 1/3.

RT @dhir_anish: Understanding causes is key to science. Finite observational data alone isn't enough. While Bayes offers a framework to de….

RT @dhir_anish: Excited to be presenting this work at #ICLR2025. Please do reach out if you are interested in a similar space!. 🗓️: Hall 3….

RT @mark_cummins: After using Deep Research for a while, I finally get the "it's just slop" complaint people have about AI art. Because I….

RT @mpshanahan: I agree with @AmandaAskell. It is a fallacy. It can be appropriate to say both a) that LLMs are next-token predictors, at a….

RT @AjdDavison: I'm recruiting a new PhD student on Gaussian Belief Propagation for Spatial AI. For this particular opportunity: 1. The fun….

RT @tolga_birdal: Within 2025, I will be hiring several PhD candidates and Postdoctoral researchers @ICComputing to work on various aspects….

RT @wooldridgemike: Simultaneously funny and very relevant question… (Answer: no it doesn’t. But what’s missing, exactly?).

Test sets are the "get out of jail free" card of ML. It allows us to apply any crazy statistically-unprincipled method, and simply check if the predictions are good. But, it's easy to forget that holdout estimates themselves rely on assumptions that can be violated in practice!.

Looking forward to #GenU2024, my favourite gathering in the AI/ML calendar. The talks again all look incredibly interesting!.

RT @KevinKaichuang: A Gaussian Process regression model built on top of protein language model embeddings and inverse folding models makes….

It was great to work with @folch_pablo, @CalvinTsay and @RuthMisener. BASF's support, collaboration, and interest in these methods contributed strongly to the success!. I look forward to our next project on experiment design and low-data decision-making. And to Jose's next steps!.

It's great to see (soon Dr) Jose Pablo Folch's PhD research recognised by a Best Paper Award🎉. Jose improved experimental design and data acquisition when experiments run in parallel, with cost/reliability tradeoffs. GPs remain key in these data-constrained industrial problems!.

The Dept of CS, and the University of Oxford more generally, also have funding available. If you apply by the December 1st deadline you will automatically be entered to all scholarships that you are eligible for. Now is a good time to start thinking about applications!.

This is a great scholarship, for those who need it most, to fund Masters/Doctorates in Maths/Stats. If you are interested in doing a DPhil/PhD with me, this scholarship can fund your studies. Interested? See some thoughts on our research and how to apply:.