Massimiliano Luca
@luca_msl
Researcher @FBK_research. Previously at @TEFresearch | @IFISC_mallorca | @medialab @unibz . Generalization and geographically-transferable mobility models
New in PNAS! 📢 lnkd.in/d7z2VEjh 👉 Combining individual & collective behavior boosts out-of-routine mobility prediction 👉 Collective behavior strongly impacts individual mobility in dense urban areas. 👉 The model is robust to severe behavioral changes
The #LLM wears Prada: Why #AI still shops in stereotypes #Shopping #researchers #dataSet #LargeLanguageModels #ArtificialIntelligence #Tech #Technology buff.ly/8Ya4z2T
Prevedere i movimenti umani è cruciale ma l’IA fatica quando usciamo dalla routine Una ricerca @FBK_research - @CNRsocial_ su @PNASNews propone algoritmo che unisce intelligenza individuale e collettiva per anticipare anche gli spostamenti “imprevedibili” magazine.fbk.eu/it/news/intell…
Guardare la città dall’alto non è solo un’esperienza suggestiva: può diventare un vero e proprio laboratorio per la ricerca urbana @luca_msl Science Ambassador e ricercatore @MobS_FBK, in un articolo su @NatCities ci porta sui tetti di Londra ◼️magazine.fbk.eu/it/news/guarda…
🌇 Rooftops, Reflections & a Different Kind of Writing As a researcher, I usually write about data. This time, I wrote about inspiration. My latest in @NatCities is an introspective piece on where my mind goes when I step onto a rooftop. 🔗 lnkd.in/drTiEwFe
🔥 Two new pre-prints are out: 👉 "Urban Safety Perception Trough the Lens of Large Multimodal Models: A Persona-based Approach" lnkd.in/ddSM3uez 👉 "Understanding Urban-Rural Disparities in Mobility Inefficiency for Colombia, Mexico, and India" lnkd.in/d56ZzWHB
Understanding Urban-Rural Disparities in Mobility Inefficiency for Colombia, Mexico, and India, finally out! Have a look to our paper: arxiv.org/abs/2503.01810 and Datavis: riccardodiclemente.com/projects/rural… thanks to Nandini Iyer, @luca_msl and thanks to @NetMob24 @Cuebiq @Connections_Lab
🚀 How much can Network Science boost AI? And vice versa? 🤔 Join HONS meets AI workshop at #NetSci2025 to explore these questions! 📢 Submit your work & be part of the discussion! hons-web.github.io/online/ with @alessiaantelmi @ManuelDileo @VincentPGrande @yllka_velaj @vins23p
🔊 New paper in @NatureHumBehav! Using large-scale mobility data, we find how businesses, amenities, and other urban places depend on each other. It reveals connections that aren’t always visible but have substantial economic impacts.
This paper from @takayabe_0505 et al. identifies dependencies in visits to points-of-interest and finds that these dependencies improve predictions of business resilience to shocks. @estebanmoro @mrfrank5790 @alex_pentland @nyutandon @NYU_CUSP @NUnetsi nature.com/articles/s4156…
🎉 Our work "SMOSE: Sparse Mixture of Shallow Experts for Interpretable Reinforcement Learning in Continuous Control Tasks" has been accepted by AAAI 2025! #AAAI2025 paper: arxiv.org/abs/2412.13053 with @L_Ferrarotti, @brulepri, @gih82 🧵1/6
🚀 New on arXiv "Towards Civic Digital Twins: Co-Designing the Citizen-Centric Future of Bologna" arxiv.org/abs/2412.06328 🛠️ Co-design & citizen engagement are key to future urban development. We explored the vision, challenges, & opportunities for Civic Digital Twins.
What an incredible experience at the European Mobility Symposium! Huge thanks to @ric_dicle for co-organizing and to all the participants for the enriching discussions. See you at the next edition! #EMoS #HumanMobility @CNRsocial_ @Northeastern @scuolanormale @NUnetsi
Complex systems like transport and energy can't be understood by looking at individual parts The Complex Systems Society promotes an interdisciplinary approach: members include FBK researchers @ricgallotti @luca_msl @KavehKadkhoda @lualvarezzuzek @t_louf shorturl.at/qamWe
➡️What can we learn from directionality in group interactions? ⬅️In “The microscale organization of directed hypergraphs” we study structural properties of systems with directed, higher-order interactions 📄Arxiv: arxiv.org/abs/2410.16258 w/ Vendramini, Montresor and @fede7j
👉 Extremely excited to share that with me and @ricgallotti, have been elected, together with other colleagues from other institutions, as members of the council of @CompSysSoc More info at cssociety.org/about/council @FBK_research @MobS_FBK @chub_fbk
Seemingly unbiased LLMs exhibit "collective biases" when interacting with other agents. We ought to look at AI biases *within application contexts*, not simply by probing models in isolation. Great collab. with @RELflintashery and @a_baronca @nerdsitu @ITUkbh @Carlsbergfondet
We explore how Large Language Models (LLMs) autonomously develop social conventions through interaction, without external programming or centralized coordination. 2/7
🌍 New personal website! It contains only personal information, but it will become a space to share project insights. Scientific outcomes often do not reach the general public, and with this space, I will do my best to bridge this gap. 👉 mluca.page
🏆 We did it again! Nice to see Nandini Iyer and @ric_dicle collecting the best data challenge paper award at @NetMob24! So sad not to be there to celebrate with them 😕 "Assessing Urban and Rural Routing Inefficiencies with Aggregated Mobility Patterns" Stay tuned for news!
#ELIAS is proud to have supported #EDS2024! 🎓Over 100 PhD students from Europe gathered in Paris to explore #AI & #sustainability, featuring expert keynotes, an industry fair, & networking sessions. Next stop: Warsaw - @IDEAS_NCBR for #EDS2025! Read more: elias-ai.eu/news/events/ed…
Mobility-GNN: a human mobility-based graph neural network for tracking and analyzing the spatial dynamics of the synthetic opioid crisis in the USA, 2013-2020 arxiv.org/abs/2409.09945