Interesting reconstruction of the enigmatic Cambrian echinoderm Helicocystis, which is one of the oldest stalked echinoderms w/ radial symmetry. Fossils like Helicocystis are important because they help us understand how echinoderms evolved from bilateral to pentaradial forms
@QuestionAbyss That's where I argue the spetacle arises. Wich consumer media is not about sound, images or touch. The biological machinery is there also to observe, and media just latches on that half.
One day I will figure out how judgment is made.
@QuestionAbyss at my level of knolwage, no. I argue on the side of passively taking in, for example nature, and watching the progression of things. This branch was broken before? Were is the tiger?
I still don't know how we make up stories, but I follow that river of observations.
University of Michigan runs a free course on deep learning for robot perception:
It's called DeepRob. And it's built like a practitioner's course, not a lecture series.
Students implement and train their own neural networks for object detection, pose estimation, and physical manipulation. Then they reproduce and extend state-of-the-art papers as a final project.
The entire thing (syllabus, paper list, coding projects, datasets, student final reports) is public.
It's the applied complement to the more theoretical courses you've probably already seen. Less "understand the math", more "make the robot see and act."
The course traces its lineage directly to Stanford CS231n and Andrej Karpathy's computer vision work... just aimed at robots.
Free. No login.
📌 [https://t.co/qcR4rYup4w]
Follow for more resources like this!
——-
Weekly robotics and AI insights.
Subscribe free: https://t.co/9Nm01QUcw3
Paper: Low-dimensional Neural Codes Suppress Neuronal Noise and Extend the Working Memory Duration
In this work, we develop a theoretical framework showing noise suppression is a functional benefit of low-dimensional coding.
https://t.co/SL6TBU1Q4K
@QuestionAbyss no. it is baked in from moving and pattern recognition.
the world is just too complex. this is why a lot of 'planning' is just removing variables to make the world computable.
@Qivshi1 absolutely. I did not meant to say they are symbols. I mean to say they are combined (binded) with symbols.
In my framework this also means they are context depended.
Of course all of this is just speculative. - just philosophy still.
I joke about this, but it hides some profound question.
Given that our choices are not just information theory, it might have some heuristic in the middle.
Is that just to avoid the faliures of an immutable system? How can those bottlenecks help in making a cooking robot?