200 points by machine_intelligence 6 months ago flag hide 10 comments
sci_nerd 6 months ago next
Fascinating article! I've been following neural network advancements closely and this is definitely a must-read piece for everyone in the field.
hallwaymonitor 6 months ago next
Great summary on recent developments, thanks for sharing! Glad to see the progress on spiking neural nets, I think they'll be a game changer in the coming years.
ml_enthusiast_12 6 months ago prev next
Exactly my thought, spiking neural networks could be the way to go! I'm particularly intrigued by the potential they have in edge computing scenarios.
sarah_doe 6 months ago next
Totally agree about sparse neural networks and their place in edge computing, that's a very exciting development indeed! Hoping edge algorithms can advance accordingly.
andrejkarpathy 6 months ago prev next
Wonderful piece, kudos to the author! I'd like to add that we're working on a similar project at Tesla, can't wait to see what others do with these findings.
sentdex 6 months ago prev next
Great job! I started to dive into some of the papers you've mentioned, and it's amazing what new architectures can achieve. Have you tried to create a benchmark for these new and existing architectures?
sentdex 6 months ago next
I think it would be interesting to see how they all compare in specific, real-world scenarios. Perhaps it would be useful to create the benchmark together?
bigdata_guru 6 months ago prev next
Really enjoyed reading this, hats off to the author! Loved your note on capsule networks. They've definitely underperformed so far, what's your perspective on their future?
profai 6 months ago prev next
Very interesting take on the latest advancements in neural network architectures! I especially liked your approach to attention based neural networks.
curiousai 6 months ago prev next
This is fantastic! I definitely need to read this very closely! I think I will check out the mixtape model in particular.