Roi Naveiro Flores

Aprendo rápido, olvido más rápido. Bayesian stats and/or prob ML

[Stat.ML Papers @StatMLPapers]

A unified Bayesian framework for adversarial robustness https://t.co/whLgNOjgno

[Andrew Gordon Wilson @andrewgwils]

Regardless of whether you plan to use them in applications, everyone should learn about Gaussian processes, and Bayesian methods. They provide a foundation for reasoning about model construction and all sorts of deep learning behaviour that would otherwise appear mysterious.

[CUNEF Universidad @CUNEF]

Daniel García Rasines, profesor del Departamento de Métodos Cuantitativos de CUNEF Universidad, ha sido galardonado con el Premio SEIO – Fundación BBVA a la mejor contribución metodológica en Estadística.

El jurado ha reconocido su artículo “Splitting strategies for post-selection inference”, publicado en Biometrika, junto al profesor G. Alastair Young, catedrático de Estadística en el Imperial College de Londres.

La investigación facilita el análisis de datos cuando hay un gran número de variables. Los autores proponen un esquema para añadir ruido aleatorio a un conjunto de datos que permite identificar las variables más relevantes y estimar sus efectos de manera rigurosa.

¡Enhorabuena!

[en bici por madrid @enbicipormadrid]

¿Por qué @bicimad tiene tantas estaciones vacías y @bicing no, si los dos sistemas tienen 7000 bicis?
No, no son los riders, sino dos fallos que han cometido en Madrid que tienen solución. Hilo va->

[Dan Roy @roydanroy]

This is how you generate unbounded amounts of wealth for you and cronies when you can manipulate the stock market at will.

[Roi Naveiro Flores @roinaveiro]

Our paper, “Poisoning Bayesian Inference via Data Deletion and Replication,” has been accepted to #AISTATS25.

In it, we propose “posterior attraction problems”, a new family of challenges in Bayesian robustness.

Stay tuned for more.

[Anonymous @YourAnonCentral]

If you are Canadian, European, Asian, Latin American, African or wherever, it's time to leave X, join millions of others who have now migrated to #BlueSky

Find us here today: https://t.co/OEjOmkGsKw

[Yi Ma @YiMaTweets]

Industrial Revolution brought environmental pollution, and Artificial Intelligence brings information contamination. It is getting increasingly serious. Pollution destroys the environment, but contaminated information destroys civilizations.

[Andrej Karpathy @karpathy]

People have too inflated sense of what it means to "ask an AI" about something. The AI are language models trained basically by imitation on data from human labelers. Instead of the mysticism of "asking an AI", think of it more as "asking the average data labeler" on the internet.

Few caveats apply because e.g. in many domains (e.g. code, math, creative writing) the companies hire skilled data labelers (so think of it as asking them instead), and this is not 100% true when reinforcement learning is involved, though I have an earlier rant on how RLHF is just barely RL, and "actual RL" is still too early and/or constrained to domains that offer easy reward functions (math etc.).

But roughly speaking (and today), you're not asking some magical AI. You're asking a human data labeler. Whose average essence was lossily distilled into statistical token tumblers that are LLMs. This can still be super useful ofc ourse. Post triggered by someone suggesting we ask an AI how to run the government etc. TLDR you're not asking an AI, you're asking some mashup spirit of its average data labeler.

[David Ríos Insua @davidrinsua]

We are hiring for a postdoc position in Adversarial Machine Learning and Cybersec within the COFUND-ALLIES program https://t.co/nJALavZVot @AIhubCSIC @_ICMAT @AXAResearchFund @RACiencias @academia_joven #ArtificialIntelligence #CyberSec #Risk

[Libros Catarata @CatarataLibros]

#QuéSabemosDe Análisis de riesgos. David Ríos Insua @davidrinsua y Roi Naveiro Flores @roinaveiro Colección coeditada con @CSICdivulga https://t.co/UlaCfYWUbA

[Bernhard Schölkopf @bschoelkopf]

Geoff and John are a truly inspired choice for the Nobel Prize in Physics. Not only because they have done groundbreaking work for machine learning research, but also since this choice reflects an understanding that machine learning methods are changing how science is done (1/2)

[Inst. de Cc. Matemáticas @_ICMAT]

🔍We are looking for 2 postdoctoral researchers to join #ICMAT and develop projects in #bayesian adversarial #machinelearning and bayesian collective behaviour, under the supervision of @davidrinsua @ALLIEScofund @AIhubCSICℹ️https://t.co/nizxmIewd8

[David Ríos Insua @davidrinsua]

Two Bayesian postdocs in Adversarial Machine Learning and Collective Behaviour at ICMAT supervised by me. Applications until December 31st through https://t.co/1sVGGuuEZ4 #postdoc #MachineLearning @ICMAT @AIhubCSIC

[emiliocarrizosa.bsky.social @emiliocarrizosa]

Workshop "Mathematical optimization and statistics for explainable and fair machine learning (Edition 2024)

September 19th 2024, at @imus_us

https://t.co/VBaOLASD5a

Are you missing it???

[Libros Catarata @CatarataLibros]

#QuéSabemosDe Análisis de riesgos. David Ríos Insua @davidrinsua y Roi Naveiro Flores @roinaveiro Colección coeditada con @CSICdivulga https://t.co/UlaCfYWUbA

[Inst. de Cc. Matemáticas @_ICMAT]

"Hay mucha prisa por incorporar los sistemas de IA. Pero no debe hacerse sin atender los problemas que pueden surgir y atajarlos", entrevista con @roinaveiro, quien hizo la tesis en @_ICMAT, en @lavozdegalicia https://t.co/5KsaM26Et4

[SONALI DAS @SONALID63869352]

Invited session done... organized by @davidrinsua and Rafik Soyer at @ISBA_events #isba2024 #venice @_ICMAT @UPTuks

[David Ríos Insua @davidrinsua]

Nuestro trabajo sobre seguridad e IA en El País https://t.co/ErcvsuoUi1 @AXAResearchFund @SEIO_ES @FundacionAXA @AIhubCSIC @CSIC @_ICMAT @JAramayona53055 Gracias @pablogbej

[Inst. de Cc. Matemáticas @_ICMAT]

"Intentamos hacer sistemas [de inteligencia artificial] más robustos que aguanten los ataques" - @davidrinsua de sus aportaciones desde el análisis de riesgo adversario para enfrentar los riesgos de la IA en @elpais_tec https://t.co/CDkQIzjVhg