Joshua Bon

Update of our Bayesian Score Calibration paper ahead of @bayescomp - Multivariate corrections - Helpful diagnostics - New examples: correcting VI and approximate inference for SDEs - New schematic

@HassanConor @MathMatize @tingtang222 😂

@MathMatize @tingtang222 @bonstats my bad

When you use ( and ) instead of \left( and \right)

@AdrienCorenflos Congrats mate!

I defended my PhD thesis on Monday, this was actually pretty fun. 9/10, recommend.

UPDATE: Raygun breaks her silence in response to her haters 👀⬇️

I'll shortly be advertising a PhD scholarship for a project that uses comp modelling, machine learning, and big data to examine the impact of working hours on well-being. Official advert to come soon, but feel free to get in touch now if you're interested.

Last preprint from my PhD thesis! Joint work with @KerrieMengersen @MiraAntonietta @liza_p_semenova @bonstats @QUTDataScience

@sp_monte_carlo is too modest to publicise his own work even when he's so generous with others'. So here it goes. arxiv.org/abs/2402.13678 studies the efficiency of a family of approximate slice samplers and relates it to the efficiency of the ideal (but not implementable) one.

Come listen to PhD candidate Antoine Luciano present our work on Insufficient Gibbs Sampling. He'll discuss how we can avoid ABC when we observe robust statistics (median, IQR, MAD, quantiles...). The seminar is online, but Parisians can attend in person @PariSanteCampus

@HassanConor @aistats_conf Gotta include the screenshot directly then! 😅

@bonstats @aistats_conf No links allowed in the one page response :)

How to respond to an AISTAT reviewer's comment about not being the first to explore a topic when our paper was preprinted in Jan 2023 and the paper by the reviewer was preprinted in June 2023, while adhering to anonymity guidelines? @aistats_conf

@HassanConor @aistats_conf Maybe: provide a link (GitHub?) to a screenshot of the arXiv page for the paper with private info scrubbed. Provide the arXiv url to the chair (to check) and state to the reviewer that you have done so.

@bonstats @aistats_conf This would be the safest way to ensure anonymity of the response, but reviewer was a borderline reject and explicitly said they would change score if convinced of novelty aside from other preprint, feels like this is relevant info for them

@DavidJWarne @JoshuaABull They’re using the Euler–Maruyama method

@JoshuaABull See my first thought was "why are they using a discrete set to enumerate Continuous Improvement Workshops?"

I don't know what this event is, but I assume it must be better than Continuous Improvement Workshops 1 & 2

Le séminaire "Mostly Monte Carlo" (un vendredi par mois à @PariSanteCampus) a maintenant sa page web et sa liste de diffusion : bonstats.github.io/mostlymontecar…

Abstract submissions close today (Tues 31 Oct) for Bayes on the Beach. We'd love to hear from you and have you join us for this conference on the sunny ☀Gold Coast in February! research.qut.edu.au/qutcds/bayes-o…

I'm running for mayor of Brisbane! I think a radical, creative campaign can help fundamentally transform unjust systems. Brissie is the tail that wags the political dog of Australian politics, and the 2024 council election is gonna be very interesting. jonathansri.com/candidacy

SMC Down Under! Thanks to all the speakers, helpers and fellow organisers. What a week! @QUTDataScience @AustMS @DiscoverAMSI

We're proud to be hosting and co-sponsoring two events at @QUT this week -#AMSIWinterSchool2023 and #SMCdownunder. Today, they got together for some workshops - and a nice combined group shot! @DiscoverAMSI @AustMS

@BAT_AUS95 🚀

@bonstats What currency is that in? And what are you trying to upgrade? A trip to Mark from economy to business? 😵‍💫😉🤣

Cursed scientific accuracy

@DavidJWarne That could be it!

@bonstats Numerical underflow?