AIXI = Solomonoff Induction ∘ Sequential Decision Theory When I say "decision tree" you should think of "sequential decision theory". Sequential decision theory relies upon Solomonoff Induction, with which it is composed into a top down theory of AGI wherein the future-discounted reward for a decision (action) is inferred based on the best unsupervised model of the available data (ie: the most highly compressed form of the data). PPM is just one technique for approximating this "best" model. Q learning, as with all reinforcement learning schemes, implicitly compresses its experiences in a quasi supervised manner by associating each state with a probability distribution of reward by their associated actions.
Hopefully I got all that right. On Mon, Aug 24, 2020 at 4:23 PM <[email protected]> wrote: > Totally lost here....can you build on text prediction (PPM, ok?) (how Q RL > learning would work/ fit in). And, whatever you're saying.... > *Artificial General Intelligence List <https://agi.topicbox.com/latest>* > / AGI / see discussions <https://agi.topicbox.com/groups/agi> + > participants <https://agi.topicbox.com/groups/agi/members> + delivery > options <https://agi.topicbox.com/groups/agi/subscription> Permalink > <https://agi.topicbox.com/groups/agi/T0291e6571222d56c-M613d7acdfe02eb4913988c3b> > ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T0291e6571222d56c-M7934acb5fe8ef25094f3d0f4 Delivery options: https://agi.topicbox.com/groups/agi/subscription
