Head of the Perception and Intelligence Lab (PINlab) and professor @ Computer Science Dept Sapienza University Research on computer vision and machine learning https://fgalasso.bitbucket.io/https://bsky.app/profile/galasso.bsky.socialhttps://bsky.app/profile/galasso.bsky.social/post/3movbe3exwc2hWe find out: the agent's policy state is enough to estimate the future action outcome! we'll present #ECCV2026 in Sweden. With our plain #LLM (dubbed DILLO), the agent finds it out 14x faster. Performance improves up to 15pp. This is #SafeAI for #agents arxiv.org/abs/2603.23149 [contains quote post or other embedded content]22 Jun 2026 16:06 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3movbe3exwc2hhttps://bsky.app/profile/galasso.bsky.social/post/3movbcm5uhc2h#Worldmodels are great! --with those, #agents can simulate the outcome of their actions, understand whether they will be successful, and safe. BUT world models are expensive, and not compatible with real-time agent's decision making. [contains quote post or other embedded content]22 Jun 2026 16:05 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3movbcm5uhc2hhttps://bsky.app/profile/galasso.bsky.social/post/3lnklkmmzrc2tEgoAct: 1st Workshop on Egocentric Perception and Action for Robot Learning @ RSS2025 egoact.github.io/rss2025/ [contains quote post or other embedded content]24 Apr 2025 11:18 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lnklkmmzrc2thttps://bsky.app/profile/galasso.bsky.social/post/3lhvpbqlazk2o"Compositional Entailment Learning for Hyperbolic Vision-Language Models" #ICLR25 has been 𝐬𝐞𝐥𝐞𝐜𝐭𝐞𝐝 𝐟𝐨𝐫 𝐚𝐧 𝐨𝐫𝐚𝐥 𝐩𝐫𝐞𝐬𝐞𝐧𝐭𝐚𝐭𝐢𝐨𝐧 🏆 1.8% of the submitted papers were selected for an oral [contains quote post or other embedded content]11 Feb 2025 12:53 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lhvpbqlazk2ohttps://bsky.app/profile/galasso.bsky.social/post/3lhvpap2eyk2o"Following the Human Thread in Social Navigation" #ICLR25 has been 𝐬𝐞𝐥𝐞𝐜𝐭𝐞𝐝 𝐚𝐬 𝐬𝐩𝐨𝐭𝐥𝐢𝐠𝐡𝐭 🏅 5.1% of the submitted papers were selected for a spotlight or an oral [contains quote post or other embedded content]11 Feb 2025 12:52 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lhvpap2eyk2ohttps://bsky.app/profile/galasso.bsky.social/post/3lgkyr2aihm2h"Following the Human Thread in Social Navigation" #ICLR25 With: Luca Scofano, Alessio Sampieri, Tommaso Campari, Valentino Sacco, Indro Spinelli, @lambertoballan.bsky.social Paper: arxiv.org/abs/2404.11327 [contains quote post or other embedded content]25 Jan 2025 13:18 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lgkyr2aihm2hhttps://bsky.app/profile/galasso.bsky.social/post/3lgkypl4lju2h"Compositional Entailment Learning for Hyperbolic Vision-Language Models" #ICLR25 With: Avik Pal, Max van Spengler, Guido Maria D'Amely di Melendugno, Alessandro Flaborea, and @pascalmettes.bsky.social Paper: arxiv.org/abs/2410.06912 [contains quote post or other embedded content]25 Jan 2025 13:17 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lgkypl4lju2hhttps://bsky.app/profile/galasso.bsky.social/post/3lgky455zp42hOur @pinlab.bsky.social has 2 #ICLR25 papers accepted! I'm proud of the strong team and of the fantastic collaborators.25 Jan 2025 13:06 +0000at://did:plc:bnltujprpkrstforsj6hbued/app.bsky.feed.post/3lgky455zp42h