Twitter | Pretraživanje | |
Rajeev
Vision, cognition, AI researcher. 🤖➕🧠➕💻➕👁 Alum of and Adventurer; Food enthusiast
148
Tweetovi
130
Pratim
81
Osobe koje vas prate
Tweetovi
Rajeev 25. pro
Amazing work, congrats !!
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Vicarious 7. pro
You'd think a generative model should answer different queries but deep learning models like VAEs are trained to answer just one type of query! We introduce a method called query-training that creates an inference network that can answer novel query types.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 1. stu
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Blake Richards 28. lis
1/ SciTwitter: I'm very excited to share our new Perspective article out in Nature Neuroscience today!
Reply Retweet Označi sa "sviđa mi se"
Rajeev 13. lis
Haha, this is brilliant
Reply Retweet Označi sa "sviđa mi se"
Rajeev 11. lis
Odgovor korisniku/ci @tyrell_turing
These guys have some data too
Reply Retweet Označi sa "sviđa mi se"
Rajeev 11. lis
Odgovor korisniku/ci @tyrell_turing
Got it. My point really was there is nothing special (statistically or spatially) that you will get from mouse cam movies, except for loads of movement artefacts, unless you want to model optic flow or smth from a rodents perspective.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 11. lis
Odgovor korisniku/ci @tyrell_turing
Basically, replace the Fourier amplitude spectrum of a natural scene with a noise pattern of your choice. Mice love em.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 10. lis
Odgovor korisniku/ci @tyrell_turing
I believe Tom Mrsic_Flogel at UCL had some mouse cam movies. Alternatively I have a method to modify the statistics of a natural movie. Published j neuro in 2015. Happy to discuss more.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 1. lis
Odgovor korisniku/ci @missippy @nyulangone i 7 ostali
congrats
Reply Retweet Označi sa "sviđa mi se"
Rajeev 15. ruj
Odgovor korisniku/ci @arisbenjamin @svoboda314
And that's not including the carbon cost of printing all these posters that will likely only be used once. an archaic practice that has to stop.
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Dileep George 13. ruj
Our 2nd paper introduces Memorize-Generalize, an algo for fast and flexible learning of sequential structure: it can rapidly memorize sequences as they are encountered, and gradually generalize as more data is accumulated.
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Dileep George 6. ruj
The preprint of my commentary on 'Building machines that work like the brain' is now on arxiv. It distinguishes between AGI & AUI and uses 'airplanes don't flap wings' as the reason to look to the brain for building
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Tamás Görbe 5. kol
A Mathematician's Way* of Converting Miles to Kilometers • 1 mi ≈ 1.609 km • The ratio of consecutive Fibonacci numbers Fₙ₊₁/Fₙ tends to the golden ratio φ ≈ 1.618 as n increases So Fₙ mi ≈ Fₙ₊₁ km *This is how I do it 😅
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Vicarious 27. kol
Our paper proposes a new model for cognitive maps in the hippocampus which involves learning higher-order graphs of event sequences. It can explain flexible planning, route-dependence of place cells, and subsumes successor representations.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 7. kol
Fantastic article, highly recommend reading.
Reply Retweet Označi sa "sviđa mi se"
Rajeev 4. kol
... And the point of this tweet is? Does the data not suggest a need to ban assault weapons. Also aren't the shootings another data point? Utter nonsense.
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
mardinly 18. srp
Odgovor korisniku/ci @AndrewHires @elonmusk i 4 ostali
like what you hear? we are hiring neuroscientists! your biological and technological distinctiveness will be added to our own
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
Nick Steinmetz 16. srp
PUBLIC SERVICE ANNOUNCEMENT: You must cross-validate sorted plots!! E.g. peak-latency sorted colormaps of neural activity. Otherwise you WILL find "sequences" out of random noise. Don't believe me? A proof in four lines. Anyone feel like you have seen the last panel before?
Reply Retweet Označi sa "sviđa mi se"
Rajeev proslijedio/la je tweet
The Dodo 6. srp
Imagine if this happened to you 🦍💗
Reply Retweet Označi sa "sviđa mi se"