iiinova — where your digital self keeps evolving

Reposted by Giovanni Saraceno

Marcelo Rinesi @marcelorinesi.bsky.social · 17/07/2026

Interesting! Moving from "*the* model" to "a handful of" to "the space of" strains culture, training, and political assumptions in science and industry but IMHO it's how we can best turn increased computational power into better research. @giovannisaraceno.bsky.social @antcalcagni.bsky.social

arxiv.org

Towards a Geometric Characterization of Multiverse Analysis

Multiverse analysis makes explicit how empirical conclusions depend on alternative, defensible analytical specifications. Standard approaches usually generate the multiverse first and then summarize…

032

Reposted by Giovanni Saraceno

Emil Hvitfeldt @emilhvitfeldt.bsky.social · 11/06/2026

Having a visual component when you are teaching a new concept is always something I value. I went back and redid some old diagrams as animations This website has all mp4, gif and code for you to use yourself in slides of websites! emilhvitfeldt.github.io/tidy-animati... #rstats #quarto #dataBS

1495

Reposted by Giovanni Saraceno

Antonino Greco @agreco.bsky.social · 17/04/2025

It was a pleasure to present my current research on cogntive modeling from ideal observer models to deep neural networks at the University of Padua @unipd.bsky.social, hosted by @giovannisaraceno.bsky.social! Lots of great feedback and scientific exchange 🙏 www.stat.unipd.it/sites/stat.u...

052

Reposted by Giovanni Saraceno

Mark Ramos @mframos.bsky.social · 05/01/2025

Trick to understanding confidence intervals (ci): "Probability is on the method; confidence is in the output." The probability statement about ci's is only true prior to data collection, but that doesn't stop you from being "confident" about your interval. #edusky #statsky #academicChatter #phdchat

3186

Reposted by Giovanni Saraceno

Ioannis Kontoyiannis @kontoyiannis.bsky.social · 16/12/2024

🤖Just saw this extremely evocative paper on deep NNs in the Bulletin of the AMS. The idea of *compositional sparsity* seems be very interesting for both theorists and practitioners. Published version: www.ams.org/journals/bul... Open-access version: cbmm.mit.edu/sites/defaul...

Title and abstract of "Compositional sparsity of learnable functions" by Poggio and Fraser

061

Reposted by Giovanni Saraceno

arxiv.stat.ME @arxiv-stat-me.bsky.social · 24/07/2024

Marianthi Markatou, Giovanni Saraceno A unified framework for multivariate two-sample and k-sample kernel-based quadratic distance goodness-of-fit tests arxiv.org/abs/2407.16374

001

Reposted by Giovanni Saraceno

arxiv.stat.ME @arxiv-stat-me.bsky.social · 10/01/2024

Claudio Agostinelli, Luca Greco, Giovanni Saraceno Weighted likelihood methods for robust fitting of wrapped models for $p$-torus data. (arXiv:2401.04686v1 [stat.ME]) arxiv.org/abs/2401.04686

001