diff --git a/docs/recognizing_categories_and_instances.md b/docs/recognizing_categories_and_instances.md index c2b6346..92ef18b 100644 --- a/docs/recognizing_categories_and_instances.md +++ b/docs/recognizing_categories_and_instances.md @@ -90,7 +90,7 @@ The optimal instance recognition algorithm, for normally distributed attributes, and for already existent, already known categories, is Mahalanobis distance -Is not the spam characteristic of an email just its $T.(S-G)$, where $T$ is +Is not the spam probability of an email just its $T.(S-G)$, where $T$ is the vector of the email, and $S$ and $G$ are the average vectors of good email and spam email? @@ -103,9 +103,18 @@ normal. But variance is, approximately, the log of probability, so Mahalanobis is more or less Bayes filtering, or at least one can be derived in terms of the other. -So we can reasonably reduce each email into twenty questions space, or, just -to be on the safe side, forty questions space. (Will have to test how many -dimensions empirically retain angles and distances) +So we can reasonably reduce each email into twenty questions space, albeit in practice, a great deal more than twenty. Finding far from random +dimensions that reduce it to a mere twenty or so is an artificial intelligence +hard problem. If random dimensions, need $\bigcirc20\log{(n)}$ dimensions +where $n$ is the number of things. And $n$ is apt to be very large. + +Finding interesting and relevant dimensions, and ignoring irrelevant and +uninteresting dimensions, is the big problem. It is the tie between +categorizing the world into natural kinds and seeing what matters in the +perceptual data while ignoring what is trivial and irrelevant. This requires +non trivial non local and non linear combinations of data, for example +adjusting the perceived colour of the apple for shadow and light colour, so +see the ample, rather than merely the light scattered by the apple into the eye. We then, in the reduced space, find natural groupings, a natural grouping being an elliptic region in high dimensional space where the density is diff --git a/docs/social_networking.md b/docs/social_networking.md index ef60daf..a5ce97c 100644 --- a/docs/social_networking.md +++ b/docs/social_networking.md @@ -363,6 +363,34 @@ way hash, so are not easily linked to who is posting in the feed. This handles public posts. +### Kademlia in social space + +The vector of each identity is a sequence ones and zeros of unbounded +length, unboundedly large dimension, but in practice you will not need +anything beyond the first few hundred. + +We deterministically generate the vector by hashing the public key of the identity. + +We do not have, and do not need, an exact consensus on how much of the +vector to actually use, but everyone needs to use roughly the same amount +as everyone else. The amount is adjusted according to what is, over time, +needed, by each identity adjusting according to circumstances, with the +result that over time the consensus adjusts to what is needed. + +Each party indicates what entities he can provide a direct link to by +publishing the sum of the vectors of the parties he can link to - and also +the sum of the their sums, and also the sum of their ... to as many deep as +turns out to be needed in practice, which is likely to two or three such +vector sums, maybe four or five. + +When someone is looking to find how to connect to an identity, he goes +through the entities he can connect to, and looks at the dot product of +their sum vectors with target vector. + +He contacts the closest entity, or a close entity, and if that does not work +out, contacts another. The closest entity will likely be able to contact +the target, or contact an entity more likely to contact the target. + ## Private messaging Private messaging is trivial. There is no end of excellent existing software diff --git a/docs/writing_and_editing_documentation.md b/docs/writing_and_editing_documentation.md index 2e817f0..cab51ac 100644 --- a/docs/writing_and_editing_documentation.md +++ b/docs/writing_and_editing_documentation.md @@ -307,13 +307,15 @@ In this table, edited in a fixed font, you are using whitespace and blank lines ### Grid tables -Allows multiline, and alignment, but visual studio does not like it, and you still have to count those spacees +Allows multiline, and alignment, but visual studio does not like it, and you still have to count those spaces +---------------+---------------+--------------------+ | Fruit | Price | Advantages | +===============+==============:+====================+ -| Bananas | $1.34 | - built-in wrapper | -| | | - bright color | +| Bananas | $1.34 | Mary had a little lamb whose fleece was white as snow, and everywhere that | +| | | Mary went the lamb was sure to go | +| | | | +| | | bright color | +---------------+---------------+--------------------+ | Oranges | $2.10 | - cures scurvy | | | | - tasty | @@ -479,7 +481,16 @@ defined by very small source code. font-weight="400" stroke-width="2" style="text-decoration:underline; cursor:pointer;" > - + + + + + - - - + d="M14 101, c40 -20, 30 -56, 54 -18 s60 15, 40 15 c -20,0 -10,-20 0,-20 q 5,0 10,10" /> A simple scalable vector graphic