Cristina Cornelio
@Cristina__C
Followers
97
Following
12
Media
15
Statuses
35
Hope to see you there! #ICML2025 #MachineLearning #AI #LLMs #Robotics #NeuroSymbolic #RAG #KnowledgeGraphs #HierarchicalPlanning
0
0
0
Paper in a nutshell: We enhance LLM-based planners to tackle long-horizon/complex robotics tasks by combining knowledge-graph-powered Retrieval-Augmented Generation (RAG) with hierarchical task decomposition, ensuring formal correctness and reliability through symbolic validation
0
0
0
This week I’ll be at #ICML2025 presenting our new paper “Hierarchical Planning for Complex Tasks with Knowledge Graph-RAG and Symbolic Verification.” 🗓 Poster session: Wed 16 Jul, 11 a.m. – 1:30 p.m. PDT 🔗 Paper: https://t.co/TlipIxkPTj 🔗 Code:
lnkd.in
This link will take you to a page that’s not on LinkedIn
2
0
0
🚀 Breaking News in AI & Math! 🚀 Our AI-Hilbert paper, features in Nature’s AI & ML Editorial Highlight! 🌟 Check it out: Paper https://t.co/CKh2W9NaJ7 Editorial Highlight https://t.co/X96UHDbPFM GitHub https://t.co/5YzlNP9Hk8.
#AI #Innovation #Discovery #NatureComms
lnkd.in
This link will take you to a page that’s not on LinkedIn
0
4
6
Happy to share that our AI-Descartes article in Nature Communications ( https://t.co/2i7sMzTHA3) has received the Pat Goldberg Memorial Best Paper Award, an annual recognition from IBM Research for outstanding papers! 🎉 #ScientificDiscovery #AI4Science
ai-descartes.github.io
A tool for Derivable Scientific Discovery
0
1
4
Check out our latest code on the AI-Hilbert GitHub repo: https://t.co/uccwsJTsgm 🚀 #AI #ScientificDiscovery
@RyanCoryWright @InverseProblems
0
0
2
AI-Hilbert handles inconsistent theory axioms by selecting relevant subsets, as seen with Einstein's time dilation formula: given high precision atomic clock measurements and both relativistic and Newtonian theories, AI-Hilbert identifies and exclusively uses the relevant one.
1
0
0
AI-Hilbert can derive symbolic expressions from a consistent background theory alone, as seen with the Hagen-Poiseuille equation. It can also handle incomplete background theories, by exploiting numerical data to compensate for missing axioms.
1
0
0
Given the background theory, the data and a set of hyperparameters, AI-Hilbert formulates scientific discovery as a polynomial optimization problem, reformulates it as a semidefinite optimization problem, and solves it obtaining both a symbolic model and its formal derivation.
1
0
0
AI-Hilbert proposes new scientific laws based on numerical data and a background theory defined as multivariate polynomial equations or inequalities.
1
1
5
Curious about AI-Hilbert? Dive into our paper here: https://t.co/EZy1Cboyz7 and explore more on our website: https://t.co/vAnMaDuMJ1
#AI #reasoning #scientificdiscovery
0
0
0
Excited to announce that our new paper, AI-Hilbert, is now published in @NatureComms! A follow-up to last year AI-Descartes, AI-Hilbert uses polynomial optimization and logical reasoning to revolutionize automated science discovery. Check it out! @RyanCoryWright @InverseProblems
1
0
4
AI-Hilbert shows that theory and data can be unified into a new scientific discovery paradigm using polynomial optimization, @RyanCoryWright @InverseProblems #GettingApplied
https://t.co/jYuXjXHaqZ
nature.com
Nature Communications - Scientific discovery is a highly relevant task in natural sciences, however generating scientifically meaningful laws and determining their consistency remains challenging....
0
2
5
Our paper "Evolving scientific discovery by unifying data and background knowledge with AI Hilbert" w @Cristina__C , Sanjeeb Dash, Bachir El Khadir, @InverseProblems has appeared in @NatureComms! It evolves the discovery phase of scientific method using polynomial optimization
1
3
11
📢Call for Papers: Submit your latest research on Neurosymbolic Generative Models for our special issue. Fast track for recent conference papers available. Deadline: Nov 15, 2024. https://t.co/puPC5Q6pLB
#AI #NeuroSymbolic #DeepLearning #GenerativeModels #NLP #reasoning #logic
0
0
2
Check out our newly accepted work at #IROS2024! ➡️ https://t.co/lFuS0QCLLa In this paper, we introduce RECOVER, a real-time #NeuroSymbolic method that combines ontologies, logical rules, and LLM planners for failure detection and recovery in #Robotics. #AI #AIReasoning #IROS
0
0
2
[INTERNSHIP POSITIONS July-December 2024] Are you an outstanding PhD student looking for an internship with a top industrial AI lab? Check out our Internship Programme at Samsung AI Centre, Cambridge! https://t.co/G0R8TRoGv6
0
0
1
🤖🔍 Check out this intriguing @arstechnica article dissecting the power of our AI-Descartes method within the realm of scientific discovery! 🧪📚 #AIReasoning #Regression #ML #AI #ScientificDiscovery #AI4science #ScientificTheories
arstechnica.com
Getting AI to find scientific laws sometimes works, but it’s a long way from science.
0
1
5
What does it take to get AI to work like a scientist? https://t.co/4pYSJ0d2B3
arstechnica.com
Getting AI to find scientific laws sometimes works, but it’s a long way from science.
3
2
5
If you want to know more about NASR (Neural Attention for Symbolic Reasoning) the recording of my presentation at #NeSy2023 is now available here:
0
1
2