On Mon, Jun 29, 2020, 4:24 PM <immortal.discover...@gmail.com> wrote:
> But isn't Occam's Razor just basically saying "simple but not too simple"? > And isn't it razorring off search space to look through when looking for a > solution? > No. A razor is a witty saying. Occam said essentially that simple explanations are better. Simple, as in fewer words to describe what you observed, and better, as in making more accurate predictions of future observations. Solomonoff, Kolmogorov, and Chaitin all formalized this to mean that shorter programs that reproduce the same data are better predictors of future data. This is the foundation of machine learning, and why we use compression to evaluate AI. The fact that there is no general procedure to find the shortest program (proved by Kolmogorov) is what makes AI (and compression) hard and what makes Hutter's AIXI algorithm for reinforcement learning both optimal and not computable. You should probably study this if you want to work in AI. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T37756381803ac879-M9a69dd0e851e1df3a81a862a Delivery options: https://agi.topicbox.com/groups/agi/subscription
