Nature as architect, man the mimic. Natural structures yield peculiar patterns. Here, in collaboration with Julie Huynh , we emulate neural networks via embodied computation.
The system utilize electrolysis in aqueous and gelatinous copper sulphate solution to fuse electrodes, creating hard wired electrical connections over time as a result of accumulated current flow. Designed to learn by creating a fusing contact of copper deposited between electrodes. The system has two states, a Learning State and a Memory State. In Learning State sensors send patterns to the memory pak via transistors for amplification thus fusing contacts over time. In Memory State the memory pak recalls via a small test current run through the fused or unfused contacts.
The system remembers patterns, and forget patterns via chemical processes and analog logic gates. The remembered patterns are utilized to stimulate output actuators.