🔥The 9th Round of Easy Loan, Earn $40 Reward is in progress❗️ ⏰ Promotion Period: January 15th - Feburary 15th, 2025 👉 Register now and check more details at gate.io/campaigns/358

Tactical Analysis of why and how Flick's Barclona won 5-2 against Real Madrid in the Super Cup final (Thread🧵) Likes and Reposts are much appreciated !

We just concluded a productive meeting of the European project #TRUST4AI, joined by colleagues from Mondragon Univ. and Universidad de Sevilla. The project has made significant strides in identifying biases and safety issues in well-known LLMs! Let's keep working towards trustworthy AI!

Early External Safety Testing of OpenAI's o3-mini Insights from the Pre-Deployment Evaluation

OpenAI’s o3-mini model is about to get released to the general user! The #trust4ai team has been part of the external safety testing group. How safe is this model? Have a look at our paper!

Interesante paper de universidades españolas que ya han tenido acceso al modelo o3-mini a través de su programa de External safety testing! Comparado con modelos anteriores o3-mini parece ser un modelo más seguro 👌

#OnlineFirst Defining and generating multi-level and uncertainty-wise test oracles for cyber-physical systems Pablo Valle, Aitor Arrieta, Liping Han, Shaukat Ali & Tao Yue doi.org/10.1007/s10270…

Our new paper “Red Teaming Contemporary AI Models: Insights from Spanish and Basque Perspectives” is on Arxiv: arxiv.org/pdf/2503.10192 We conduct a red teaming of 3 SOTA LLMs in Basque and Spanish, with the goal of testing their safety and bias!

On Tuesday, Pablo Valle will attend the PhD symposium and present his work “Automated Repair of Cyber-Physical Systems”. His poster will also be available during the main conference from Wednesday to Friday at 17:30.

Mondragon team in action yesterday and today at ICSE

Don’t miss today at 2 PM the talk by Pablo Valle on “MarMot: Metamorphic Runtime Monitoring of Autonomous Driving Systems”

Time for Pablo Valle to present “ MarMot: Metamorphic Runtime Monitoring of Autonomous Driving Systems”, accepted in tosem!

🤩 𝐈𝐤𝐚𝐫𝐚𝐠𝐚𝐫𝐫𝐢𝐚𝐚𝐚! 🙌🏼 El Eibar femenino C logra el ascenso a 3ªRFEF tras ganar 3-0 al Bizkerre B, a falta de 2 jornadas. 𝐙𝐎𝐑𝐈𝐎𝐍𝐀𝐊 𝐧𝐞𝐬𝐤𝐚𝐤! ❤️💙 #BetiArmaginak ⚔️

After a great week at ICSE, the week couldn’t begin better! Our paper “An Industrial Experience Report on Applying Search-based Boundary Input Generation to Cyber-Physical Systems” accepted at emsejournal! Work led by Pablo Valle, together with Vincenzo Riccio and Paolo Tonella

It is a pleasure to work together and learn from such experienced researchers! I am very proud of this work

😍 Omenaldi berezia! 🔝 Gure emakumezkoen C taldearen agurra, aurreko astean lortutako 3ªRFEF mailara igoeragatik. #EibarFem | #BetiArmaginak ⚔️

Evaluating Uncertainty and Quality of Visual Language Action-enabled Robots. arxiv.org/abs/2507.17049

A few days ago, I saw Tesla’s Optimus humanoid robot serving some popcorns in their new “Tesla Diner” restaurant. After digging a bit into it, I found that apparently, this was teleoperated.

Our new study tackles this critical blind spot in VLAs for robotic systems 👇 🚀 Just published on arXiv: "Evaluating Uncertainty and Quality of Visual Language Action-enabled Robots" 📄 arxiv.org/abs/2507.17049

📖 Full paper: arxiv.org/abs/2507.17049 Co-authored with Pablo Valle, Chengjie Lu, Shaukat Ali #Robotics #TrustworthyAI #VLA #MachineLearning #AIEvaluation #AutonomousSystems #VisionLanguageModels #UncertaintyEstimation #EmbodiedAI #SoftwareEngineering

Interested in quantifying uncertainty or measuring the quality of Visual Language Action (VLA) models? Have a look to our new paper! #VLA #Robotics #trustworthyAI #embodiedAI