Shane Gero @[email protected]

4.3K posts

Shane Gero @sgero@scicomm.xyz banner
Shane Gero @sgero@scicomm.xyz

Shane Gero @[email protected]

@sgero

Canadian Biologist @CarletonBiology @DalScience & @AarhusUni | @natgeoexplorers | Biology Lead @ProjectCETI | Founder @DomWhale | He/Him

Canada Katılım Nisan 2009
1.4K Takip Edilen2.2K Takipçiler
Shane Gero @[email protected] retweetledi
Financial Times
Financial Times@FT·
The same technology that is powering ChatGPT could allow us to reach into the non-human world, and begin to understand — even speak — animal languages, or so researchers hope. Here’s how: on.ft.com/4aYsmU5
English
8
26
64
65K
ElBeth
ElBeth@lisabethsparkle·
@sgero I can picture your children’s book in my head.
English
1
0
1
6
Shane Gero @[email protected] retweetledi
Bioneers
Bioneers@bioneers·
Until fairly recently, the dominant view among scientists was that non-human animals didn’t manifest real intelligence or live in dynamic cultures. Read how @carlsafina and @sgero are challenging this theory in a conversation with journalist @meownderthal bit.ly/42kA31l
Bioneers tweet media
English
0
4
9
752
Shane Gero @[email protected] retweetledi
Gašper Beguš
Gašper Beguš@begusgasper·
Our network learns properties considered meaningful by existing research, such as: -Number of clicks -ICI timing The network also uncovers new acoustic properties that may be meaningful but haven't been hypothesized as such before: -Spectral mean -Spectral regularity
English
2
4
49
4.4K
Shane Gero @[email protected] retweetledi
Gašper Beguš
Gašper Beguš@begusgasper·
How do we approach an unknown communication system? We train GANs that learn to imitate whale vocalizations and encode information into those vocalizations. Using CDEV (causal disentanglement with extreme values), we can infer which properties the model encodes as meaningful.
Gašper Beguš tweet media
English
3
5
55
5.4K
Shane Gero @[email protected] retweetledi
Gašper Beguš
Gašper Beguš@begusgasper·
Uncovering the unknowns in sperm whale communication🐳 using deep learning We introduce a novel interpretability technique (CDEV) that combines generative deep models and causal inference methods. Our model discovers meaningful properties, both novel and previously hypothesized
Gašper Beguš tweet media
English
15
113
494
167.6K