←
Giovanni Petri4,327 posts
Giovanni Petri avatar

Giovanni Petri

@lordgrilo

Topology, complex networks, neuroscience; Professor @NUnetsi; PI @NPLab_; PI @ProjectCETI; Science Comm Fellow @museumofscience; Angoleiro, wine-drinker.

◷ Joined October 2009▧ 430 Photos and videos
Surya Ganguli avatar
⟳ Giovanni Petri retweeted
Surya Ganguli@SuryaGanguli·Oct 12

Yes - reading the essay “More is different” by Anderson, which outlined the philosophy of studying complex physical systems as a distinct discipline, is useful for thinking about studying complex AI systems.

read this great essay referred to by @SuryaGanguli in one of the COLM panels: More Is Different by P. W. Anderson "The ability to reduce everything to simple fundamental laws does not imply the ability to start from those laws and reconstruct the universe." cse-robotics.engr.tamu.edu/d…

Giovanni Petri avatar
Giovanni Petri@lordgrilo·Oct 6

This! 👇👇👇👇👇👇👇👇👇

🧠🎬 What can different views of brain activity tell us about how emotion unfolds? Across 14 films (~2.5h), we compared regional activity, pairwise connectivity & higher-order topology with 50 continuously rated affective features. Our preprint is out👇 shorturl.at/AfmGv

Post attachment
Andrea Santoro avatar
⟳ Giovanni Petri retweeted
Andrea Santoro@andreasantor0·Oct 6

🧠🎬 What can different views of brain activity tell us about how emotion unfolds? Across 14 films (~2.5h), we compared regional activity, pairwise connectivity & higher-order topology with 50 continuously rated affective features. Our preprint is out👇 shorturl.at/AfmGv

Post attachment
Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 31

I’m hiring a postdoc in Dublin's DII! Can explicit topology/higher-order structure make LLMs reason better about sci/tech innovation? join us for TOPO-LLM: TDA + higher-order networks + transformers, with access to MareNostrum 5. Apply by 9 Sept: jobs.ac.uk/job/DSS062

Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 20

Very proud of this one :)

A new study by CETI Scientists analyzed thousands of hunting attempts and found that sperm whales flexibly switch prey-capture tactics within a single dive — adapting their approach in real time based on how their hunt unfolds. Read the pre-print here: bit.ly/45cW2Lz

Post attachment
Project CETI avatar
⟳ Giovanni Petri retweeted
Project CETI@ProjectCETI·Aug 20

A new study by CETI Scientists analyzed thousands of hunting attempts and found that sperm whales flexibly switch prey-capture tactics within a single dive — adapting their approach in real time based on how their hunt unfolds. Read the pre-print here: bit.ly/45cW2Lz

Post attachment
Earl K. Miller avatar
⟳ Giovanni Petri retweeted
Earl K. Miller@MillerLabMIT·Aug 20

Neural oscillations make networks work better. "Rhythm-SNN" matched or beat other networks while using over 100 times less energy. nature.com/articles/s41467-0… #neuroscience

Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 5

9/9 The larger picture: sperm whales pair persistent search with flexible, moment-to-moment hunting decisions. A testable question follows: could noise disrupt high-investment pursuits more than lower-investment captures?

Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 5

8/9 The contrast reappeared independently: Azorean echograms linked tactics to prey depth, while 2024 Project CETI tags in Dominica recovered the core kinematics. The validation samples are modest, but the convergence is encouraging.

Post attachment
Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 5

7/9 The tactics also formed sequences. Whales often repeated one, but switched roughly every 2–3 capture attempts. Type B was about twice as likely to end a dive, consistent with tactical reassessment as encounter conditions and energetic demands change.

Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 5

6/9 The contrast was substantial. Our energy model estimated that Type B attempts required ~7× more energy than Type A. This is a comparative model estimate—not a direct metabolic measurement—but the ranking held in every whale.

Post attachment
Giovanni Petri avatar
Giovanni Petri@lordgrilo·Aug 5

5/9 Zoom in on 3,686 capture attempts and two recurrent tactics emerge: A: slower, more oscillatory, with short acoustic ramp-ups. B: faster, more vertically persistent, with greater roll and longer ramp-ups. Every sampled whale used both.

Post attachment