On Thu, May 9, 2024 at 2:15 AM Rob Freeman <[email protected]> wrote:
> On Thu, May 9, 2024 at 6:15 AM James Bowery <[email protected]> wrote: > ... > Criticisms are welcome. But just saying, oh, but hey look at my idea > instead... > I may have confused you by conflating two levels of abstraction -- only one of which is "my idea" (which isn't my idea at all but merely an idea that has been around forever without garnering the attention it deserves): 1) Abstract grammar as a prior. 2) The proper structure for incorporating priors, whatever they may be. Forget about #1. That was just an example -- a conjecture if you will -- that I found appealing as an under-appreciated prior but distracted from the much more important point of #2 which was about priors in general. #2 is exemplified by the link I provided to physics informed machine learning <https://www.youtube.com/playlist?list=PLMrJAkhIeNNQ0BaKuBKY43k4xMo6NSbBa> which is appropriate to bring up in the context of this particular post about the ir/relevance of physics. The point is not "physics". Physics is merely one knowledge domain that, because it is "hard", is useful because the technique of incorporating its priors into machine learning is exemplary. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/Teaac2c1a9c4f4ce3-M4a92b688c0804deb6a6a12a1 Delivery options: https://agi.topicbox.com/groups/agi/subscription
