A Talk, led by Yasi Perera and Ezra Buchla.
“I’m going to try to make a black hole, and if you’ve ever loved someone you’ll want to hold their hand.” Not in a scary way, but I just really wanted to get down to the real huge distances between us all (intelligences), to open a huge space, and then talk about trying to make contact in that.”
Plotting out each element’s weights in a learning/self-organizing/adaptive array. The arrays have been trained on info with higher dimensions of variation than the array’s dimensionality, so they struggle to “represent” the information somewhere… the 1-d array (the points are connected to show their topological relation in the adaptive network), you can see if spreadign out in a space-filling curve-type shape.. if you plot it in 3d you get something like a 3d spacefilling curve.. not rigorously defined, just an emergent version.. an interesting aside is that as the learning progresses the curve seems to follow the iterations of a iteratively defined spacefiller like peano or whoever.. 2d array get there’s “deviation stripes” the deviation from a regular grid is plotted in the z axis. The striping formation is characteristic of biological 2d arrays of neurons which are trying to deal with higher-d input.