Similarly we inhibit questions when see the outcome we wanted, which is like cells build/branching during organ creation/regeneration.
notice the link says
"The biggest puzzle in this field is the question of how the cell collective knows what to build and when to stop."
Based on context and hardcoded desires, it grows its way forward.
"While we know of many genes that are required for the process of regeneration, we still do not know the algorithm that is sufficient for cells to know how to build or remodel complex organs to a very specific anatomical end-goal"
“build an eye hereâ€
"Imagine if we could design systems of the same plasticity and robustness as biological life: structures and machines that could grow and repair themselves."
"We will focus on Cellular Automata models as a roadmap for the effort of identifying cell-level rules which give rise to complex, regenerative behavior of the collective. CAs typically consist of a grid of cells being iteratively updated, with the same set of rules being applied to each cell at every step. The new state of a cell depends only on the states of the few cells in its immediate neighborhood. Despite their apparent simplicity, CAs often demonstrate rich, interesting behaviours, and have a long history of being applied to modeling biological phenomena."
"Typical cellular automata update all cells simultaneously. This implies the existence of a global clock, synchronizing all cells. Relying on global synchronisation is not something one expects from a self-organising system. We relax this requirement by assuming that each cell performs an update independently, waiting for a random time interval between updates"
Both local, and global shape of the context (what, and where (position)) affect the prediction.
"We can see that different training runs can lead to models with drastically different long term behaviours. Some tend to die out, some don’t seem to know how to stop growing, but some happen to be almost stable! How can we steer the training towards producing persistent patterns all the time?"
Sounds like GPT-2. When to finish a discovery sentence. Keep on topic until reach goal.
"we wanted the system to evolve from the seed pattern to the target pattern - a trajectory which we achieved in Experiment 1. Now, we want to avoid the instability we observed - which in our dynamical system metaphor consists of making the target pattern an attractor."
"Intuitively we claim that with longer time intervals and several applications of loss, the model is more likely to create an attractor for the target shape, as we iteratively mold the dynamics to return to the target pattern from wherever the system has decided to venture. However, longer time periods substantially increase the training time and more importantly, the memory requirements, given that the entire episode’s intermediate activations must be stored in memory for a backwards-pass to occur."
That sounds like Hutter Prize compressor improvement. Takes more RAM, takes longer, for better regeneration to target from Nothing (seed, compressed state).
"it’s been found that the target morphology is not hard coded by the DNA, but is maintained by a physiological circuit that stores a setpoint for this anatomical homeostasis"
We want to regenerate shape (sort the words/articles), and grow the organism/sentence as well. But avoid non-stop growth past the matured rest state goal and stop de-generation.
