iiinova — where your digital self keeps evolving

Want interpretable models whose concepts can actually be verified? Then come chat with us at ICML poster session 8 (Hall A, Thursday 17:00 to 18:45)! (4/5)

100

This makes it possible to: ✅ Inspect what each concept visually represents ✅ Spot semantic misalignment ✅ Intervene directly at the prototype level (3/5)

Concept Bottleneck Models make predictions through human-understandable concepts, but provide no way to verify whether the learned concepts match human intent. PGCMs ground each concept in visual prototypes: image parts that show the model’s evidence for a concept. (2/5)

100

David Debot @daviddebot.bsky.social · 22/05/2026

🚨 Can we trust that the “concepts” in concept-based models actually mean what we think they mean? Our new work, Prototype-Grounded Concept Models (PGCMs), makes concept alignment directly inspectable and correctable. #ICML26 Paper: arxiv.org/abs/2604.16076 🧵 (1/5)

131

Our accepted ICML paper on prototype-based Concept Bottleneck Models is now on arXiv! We introduce Prototype-Grounded Concept Models (PGCMs), enabling verifiable concept alignment through interpretable visual prototypes 🔍🧠. Check it out at: arxiv.org/abs/2604.16076

152

Reposted by David Debot

lennertds.bsky.social @lennertds.bsky.social · 24/02/2026

If you care about enforcing constraints over time without breaking your computational resources, then read our new blog post over at @aihub.org! It focuses on showing how our neurosymbolic Markov models beat the SOTA in out-of-distribution generalisation and so much more.

0104

Reposted by David Debot

Jaron Maene @jjcmoon.bsky.social · 05/12/2025

1/5 Tomorrow I’ll talk about the 𝐩𝐫𝐨𝐛𝐚𝐛𝐢𝐥𝐢𝐬𝐭𝐢𝐜 𝐩𝐫𝐨𝐠𝐫𝐚𝐦𝐦𝐢𝐧𝐠 𝐬𝐞𝐦𝐚𝐧𝐭𝐢𝐜𝐬 𝐨𝐟 𝐝𝐢𝐟𝐟𝐞𝐫𝐞𝐧𝐭𝐢𝐚𝐛𝐥𝐞 𝐩𝐫𝐨𝐯𝐢𝐧𝐠 at #NeurIPS San Diego (poster #614 11am). 📃 openreview.net/pdf?id=rEUbD... 📺 www.youtube.com/watch?v=sOTX...

openreview.net

1102

Reposted by David Debot

lennertds.bsky.social @lennertds.bsky.social · 14/05/2025

Just under 10 days left to submit your latest endeavours in #tractable probabilistic models! Join us at TPM @auai.org #UAI2025 and show how to build #neurosymbolic / #probabilistic AI that is both fast and trustworthy!

0119

Reposted by David Debot

Jaron Maene @jjcmoon.bsky.social · 23/04/2025

We developed a library to make logical reasoning embarrasingly parallel on the GPU. For those at ICLR 🇸🇬: you can get the juicy details tomorrow (poster #414 at 15:00). Hope to see you there!

1257

Reposted by David Debot

Gabriele Venturato @gabventurato.bsky.social · 28/02/2025

If you're at #AAAI2025, come check out our demo on neurosymbolic reinforcement learning with probabilistic logic shields 🤖 Tomorrow (Sat, March 1) from 12:30–2:30 PM during the poster session 💻

051

Reposted by David Debot

Jaron Maene @jjcmoon.bsky.social · 27/02/2025

We all know backpropagation can calculate gradients, but it can do much more than that! Come to my #AAAI2025 oral tomorrow (11:45, Room 119B) to learn more.

12710

Reposted by David Debot

Gabriele Venturato @gabventurato.bsky.social · 25/02/2025

🔥 Can AI reason over time while following logical rules in relational domains? We will present Relational Neurosymbolic Markov Models (NeSy-MMs) next week at #AAAI2025! 🎉 📜 Paper: arxiv.org/pdf/2412.13023 💻 Code: github.com/ML-KULeuven/... 🧵⬇️

12511

Open-source & easy to use! 🔷 Code: github.com/ML-KULeuven/... 🔷 Based on MiniHack & Stable Baselines3 🔷 Define new shields in just a few lines of code! 🚀 Let’s make RL safer & smarter, together! (7/8)

100

Want to try it yourself? 🎮 Use our interactive web demo! 🔷 Modify environments (add lava, monsters!) 🔷 Test shielded vs. non-shielded agents 🖥️ Play with it here: dtai.cs.kuleuven.be/projects/nes... (6/8)

100

Why does this matter? 🔷 Faster training ⌛ 🔷 Safer exploration 🔒 🔷 Better generalization 🌍 (5/8)

100

How does it work? 🤔🛡️ The shield: ✅ Exploits symbolic data from sensors 🌍 ✅ Uses logical rules 📜 ✅ Prevents unsafe actions 🚫 ✅ Still allows flexible learning 🤖 A perfect blend of symbolic reasoning & deep learning! (4/8)

Enter MiniHack, our demo's testing ground! 🏰🗡️ There, RL agents face: ✅ Lava cliffs & slippery floors ✅ Chasing monsters ✅ Locked doors needing keys Findings: 🔷 Standard RL struggles to find an optimal, safe policy. 🔷 Shielded RL agents stay safe & learn faster! (3/8)

100

Deep RL is powerful, but... ⚠️ It can take dangerous actions ⚠️ It lacks safety guarantees ⚠️ It struggles with real-world constraints Yang et al.'s probabilistic logic shields fix this, enforcing safety without breaking learning efficiency! 🚀 (2/8)

100

🚀 Do you care about safe AI? Do you want RL agents that are both smart & trustworthy? At #AAAI2025, we present our demo for neurosymbolic RL—combining deep learning with probabilistic logic shields for safer, interpretable AI in complex environments. 🏰🔥 🧵👇 (1/8)

184

Or check out our Medium post: 👉 medium.com/@pyc.devteam... (7/7)

020

NeurIPS Poster Interpretable Concept-Based Memory ReasoningNeurIPS 2024

With CMR, we’re reaching the sweet spot of accuracy and interpretability. Check it out at our poster at #NeurIPS2024! 👉 neurips.cc/virtual/2024... (6/7)

neurips.cc

130

During training, CMR learns embeddings as latent representations of logic rules, and a neural rule selector identifies the most relevant rule for each instance. Due to a clever factorization and rule selector, inference is linear in the number of concepts and rules. (5/7)

CMR makes a prediction in 3 steps: 1) Predict concepts from the input 2) Neurally select a rule from a memory of learned logic rules ➨ Accuracy 3) Evaluate the selected rule with the concepts to make a final prediction ➨ Interpretability (4/7)

CMR has: ⚡ State-of-the-art accuracy that rivals black-box models 🚀 Pure probabilistic semantics with linear-time exact inference 👁️ Transparent decision-making so human users can interpret model behavior 🛡️ Pre-deployment verifiability of model properties (3/7)

CMR is our latest neurosymbolic concept-based model. A proven 𝘶𝘯𝘪𝘷𝘦𝘳𝘴𝘢𝘭 𝘣𝘪𝘯𝘢𝘳𝘺 𝘤𝘭𝘢𝘴𝘴𝘪𝘧𝘪𝘦𝘳 irrespective of the concept set, CMR achieves near-black-box accuracy by combining 𝗿𝘂𝗹𝗲 𝗹𝗲𝗮𝗿𝗻𝗶𝗻𝗴 and 𝗻𝗲𝘂𝗿𝗮𝗹 𝗿𝘂𝗹𝗲 𝘀𝗲𝗹𝗲𝗰𝘁𝗶𝗼𝗻! (2/7)