digitado

About digitado

https://www.digitado.com.br

Posts by :

Why AI Products Lose Users After the First Try — And How to Design for Adoption

⋅

Most AI products don’t fail because the model is bad. They fail because the product decisions around the model are wrong — or absent. Activation looks healthy in week one, then collapses in week two. Generation counts climb while export and approval rates stay flat. Users try the AI feature once, mark it as “interesting,” and never come back. If you’re shipping AI products in 2026, you’ve seen this pattern. The question is what to do about it. […]

Ver mais

Liked Liked
La soberanía digital deja de ser una abstracción

La soberanía digital deja de ser una abstracción

⋅

Mi columna de esta semana en Invertia se titula «Europa descubre el botón de apagado» (pdf), y trata sobre cómo Europa ha empezado, por fin, a comprender que su dependencia tecnológica de las grandes compañías estadounidenses no era simplemente una cuestión de comodidad o eficiencia, sino un problema estratégico de primer orden. El detonante de esa toma de conciencia lo analiza muy bien un reciente artículo en Wired, «The EU Is going through a Trump-fueled breakup with Big […]

Ver mais

Liked Liked

Transformers Can Learn Posterior Predictive Distributions In-Context

⋅

arXiv:2605.26713v1 Announce Type: new Abstract: Prior-data fitted networks (PFNs) have recently emerged as a powerful approach for Bayesian prediction tasks, approximating the posterior predictive distribution (PPD) through in-context learning. Despite their strong empirical performance and ability to go beyond point predictions, theoretical understandings of the algorithmic capability of transformers to learn distributions in context are still lacking. Focusing on Gaussian process regression problems, we show by construction that transformers can implement a gradient descent algorithm targeting the posterior […]

Ver mais

Liked Liked

CART Random Forests as Sequential Allocation over Random Opportunity Sets: A Stochastic-Control Theory of Ensemble Risk

⋅

arXiv:2605.26675v1 Announce Type: new Abstract: CART random forests are among the most widely used modern predictive methods, with well-documented empirical success. Yet, at the mechanistic level, the algorithm is often treated as a black box because of its complexity. In this paper, we develop a stochastic-control perspective on feature-subsampled CART random forests, named CART random opportunity-set allocation (CART-ROSA). At each node, the random subset of features is interpreted as a random feasible action set, and the CART split […]

Ver mais

Liked Liked

When Does LeJEPA Learn a World Model?

⋅

arXiv:2605.26379v1 Announce Type: new Abstract: A representation that scrambles the true degrees of freedom of the world cannot support reliable planning or compositional generalization. We prove that LeJEPA (alignment plus Gaussian regularization) linearly recovers the world’s latent variables from nonlinear observations, a property known as linear identifiability, in a broad class of worlds where latents evolve under stationary, additive-noise transitions. Our main result is that among all such worlds, the Gaussian is the unique latent distribution for which […]

Ver mais

Liked Liked

Beyond Differences: Doubly Robust Meta-Learners for Ratio-Based Treatment Effects

⋅

arXiv:2605.26288v1 Announce Type: new Abstract: When treatment effects are naturally expressed as ratios — as in medicine, pricing, and marketing — the ratio-based CATE $tau(x) = E[Y|W=1,X=x] / E[Y|W=0,X=x]$ is the appropriate estimand. Yet existing estimators either impose a log-linear parametric structure or apply generic regression without robustness guarantees for this functional. We introduce the Q-Learner, which decomposes $tau(x)$ into a product of two odds ratios, reducing ratio-CATE estimation for binary outcomes to two propensity classification tasks. We […]

Ver mais

Liked Liked

Load More

↑

Digitado © 2026