announcing an annual pattern recognition competition.
the three competitors are google, microsoft, and universal structures.

the pattern recognition competition will be only on binary strings,
with 100 ordered presubmitted benchmarks by the three entrants.
the competition will be on all presubmitted benchmarks plus an
ordered set of 1000 more by each entrant at the time of competition.
the entrants must recognize all patterns correctly or be penalized
one point for each and every failure to correctly locate a pattern
occurrence. all valid benchmarks are counted. the winner of a
benchmark has the fewest memory references into the test stream
(generally called the text) and scores ten points. second place
scores five points. third place scores one point. all other places
score one point. the time limit for the competition is ten minutes
for each entrant, judged by the creation and last modified time of
the log file. plus one point for every second less than ten minutes.
minus one point for every second more than ten minutes, for all
entrants.

to verify and review correctness of all answers, a log file
containing the pattern in base ten, a comma, the text in base ten,
a comma, every occurrence location separated by commas, in base ten,
followed by a period, then a newline.

any university may enter the competition, by presubmitting ten benchmarks.
any nation may enter the competition, by presubmitting ten benchmarks.
any company may enter the competition, by presubmitting ten benchmarks.
any individual may enter the competition.

On 6/21/2021 4:59 PM, Daniel Pehoushek wrote:
> announcing an annual pattern recognition competition.
> the three competitors are google, microsoft, and universal structures.
>
> the pattern recognition competition will be only on binary strings,
> with 100 ordered presubmitted benchmarks by the three entrants.
> the competition will be on all presubmitted benchmarks plus an
> ordered set of 1000 more by each entrant at the time of competition.
> the entrants must recognize all patterns correctly or be penalized
> one point for each and every failure to correctly locate a pattern
> occurrence. all valid benchmarks are counted. the winner of a
> benchmark has the fewest memory references into the test stream
> (generally called the text) and scores ten points. second place
> scores five points. third place scores one point. all other places
> score one point. the time limit for the competition is ten minutes
> for each entrant, judged by the creation and last modified time of
> the log file. plus one point for every second less than ten minutes.
> minus one point for every second more than ten minutes, for all
> entrants.
>
> to verify and review correctness of all answers, a log file
> containing the pattern in base ten, a comma, the text in base ten,
> a comma, every occurrence location separated by commas, in base ten,
> followed by a period, then a newline.
>
> any university may enter the competition, by presubmitting ten benchmarks.
> any nation may enter the competition, by presubmitting ten benchmarks.
> any company may enter the competition, by presubmitting ten benchmarks.
> any individual may enter the competition.
>

No URL?
No hits on Google: "pattern recognition competition" ?

--
Copyright 2021 Pete Olcott

"Great spirits have always encountered violent opposition from mediocre
minds." Einstein

square of reason: from diagonal of reason

first take a look at the diagonal of reason in another thread.
thirty two 32 bit strings of binary, with a zero diagonal.
that is now called the square of reason.

we will use the square to validate pattern recognition algorithms.
search for numbers (in binary) zero through fifteen in the square.
any correct entry must find 3759 occurrences of patterns.

after validation, the entries just must answer with the number of
occurrences of the pattern in the stream or text.

one hundred instances such as:
m n pattern text recognition alphabet size two
small m, large n, ten instances, purest random bits
m=8 n=8*8*8*8*8*8*8*8 = 16777216, purest of random bits, ten instances
m=16 n=50000000, ten instances, purest random bits
m=32 n=50000000, ten instances, purest random bits
m=64 n=50000000, ten instances, purest random bits
m=128 n=50000000, ten instances, purest random bits
m=256 n=50000000, ten instances, purest random bits
m=512 n=50000000, ten instances, purest random bits
m=4 n=50000000, ten instances, purest random bits
m=2 n=50000000, ten instances, purest random bits
m=1 n=50000000, ten instances, purest random bits

pattern recognition is also identity recognition
very useful generally thoughtful area, eg cybersecurity

small m, large n, ten instances, purest random bits
m=8 n=1000000000, purest of random bits, ten instances
m=16 n=5000000000, purest of random bits, ten instances
m=32 n=5000000000, purest of random bits, ten instances
m=64 n=5000000000, purest of random bits, ten instances
m=128 n=500000000, purest of random bits, ten instances
m=256 n=500000000, purest of random bits, ten instances
m=512 n=500000000, purest of random bits, ten instances
m=4 n=5000000000, purest of random bits, ten instances
m=2 n=5000000000, purest of random bits, ten instances
m=1 n=5000000000, purest of random bits, ten instances

dear logicians please study logs

proof:

log base two of N equals stringlength of N in base two

log equals stringlength

quantumlog equals alphabetsize of string

alphabetsize equals twentyseven

log base three of twentyseven equals three

quantumlog equals log base three

quantumlog is for optimal information theory

log base three of N for optimal information theory

log base three of N for optimal information

log base three of N is optimal information

log base three of N equals optimal information

announcing an annual pattern recognition competition.
the three competitors are google, microsoft, and universal structures.

the pattern recognition competition will be only on binary strings,
with 100 ordered presubmitted benchmarks by the three entrants.
the competition will be on all presubmitted benchmarks plus an
ordered set of 1000 more by each entrant at the time of competition.
the entrants must recognize all patterns correctly or be penalized
one point for each and every failure to correctly locate a pattern
occurrence. all valid benchmarks are counted. the winner of a
benchmark has the fewest memory references into the test stream
(generally called the text) and scores ten points. second place
scores five points. third place scores one point. all other places
score one point. the time limit for the competition is ten minutes
for each entrant, judged by the creation and last modified time of
the log file. plus one point for every second less than ten minutes.
minus one point for every second more than ten minutes, for all
entrants.

to verify and review correctness of all answers, a log file
containing the pattern in base ten, a comma, the text in base ten,
a comma, every occurrence location separated by commas, in base ten,
followed by a period, then a newline.

any university may enter the competition, by presubmitting ten benchmarks.
any nation may enter the competition, by presubmitting ten benchmarks.
any company may enter the competition, by presubmitting ten benchmarks.
any individual may enter the competition.

100 pattern recognition problems

preliminary to competing, any pattern recognition must search for zero to fifteen inside the diagonal of reason and find 3759 occurrences.
the numbers zero through 15 (in binary) searched for inside the square of reason: 3759 occurrences
pattern 0 1 10 11 100 101 110 111 1000 1001 1010 1011 1100 1101 1110 1111
stream or text
11111111111111111111111111111110 1 31 1 30 0 0 1 29 0 0 0 0 0 0 1 28
11111111111111111111111111111101 1 31 1 29 0 0 1 28 0 0 0 0 0 1 1 27
11111111111111111111111111111011 1 31 1 28+1 0 0 1 27 0 0 0 0 0 1 1 26
11111111111111111111111111110111 1 31 1 27+2 0 0 1 26+1 0 0 0 0 0 1 1 25
11111111111111111111111111101111 1 31 1 26+3 0 0 1 25+2 0 0 0 0 0 1 1 24+1
11111111111111111111111111011111 1 31 1 25+4 0 0 1 24+3 0 0 0 0 0 1 1 23+2
11111111111111111111111110111111 1 31 1 24+5 0 0 1 23+4 0 0 0 0 0 1 1 22+3
11111111111111111111111101111111 1 31 1 23+6 0 0 1 22+5 0 0 0 0 0 1 1 21+4
11111111111111111111111011111111 1 31 1 22+7 0 0 1 21+6 0 0 0 0 0 1 1 20+5
11111111111111111111110111111111 1 31 1 21+8 0 0 1 20+7 0 0 0 0 0 1 1 19+6
11111111111111111111101111111111 1 31 1 20+9 0 0 1 19+8 0 0 0 0 0 1 1 18+7
11111111111111111111011111111111 1 31 1 19+10 0 0 1 18+9 0 0 0 0 0 1 1 17+8
11111111111111111110111111111111 1 31 1 18+11 0 0 1 17+10 0 0 0 0 0 1 1 16+9
11111111111111111101111111111111 1 31 1 17+12 0 0 1 16+11 0 0 0 0 0 1 1 15+10
11111111111111111011111111111111 1 31 1 16+13 0 0 1 15+12 0 0 0 0 0 1 1 14+11
11111111111111110111111111111111 1 31 1 14+15 0 0 1 14+13 0 0 0 0 0 1 1 13+12
11111111111111101111111111111111 1 31 1 15+14 0 0 1 13+14 0 0 0 0 0 1 1 12+13
11111111111111011111111111111111 1 31 1 16+13 0 0 1 12+15 0 0 0 0 0 1 1 11+14
11111111111110111111111111111111 1 31 1 17+12 0 0 1 11+16 0 0 0 0 0 1 1 10+15
11111111111101111111111111111111 1 31 1 18+11 0 0 1 10+17 0 0 0 0 0 1 1 9+16
11111111111011111111111111111111 1 31 1 19+10 0 0 1 9+18 0 0 0 0 0 1 1 8+17
11111111110111111111111111111111 1 31 1 20+9 0 0 1 8+19 0 0 0 0 0 1 1 7+18
11111111101111111111111111111111 1 31 1 21+8 0 0 1 7+20 0 0 0 0 0 1 1 6+19
11111111011111111111111111111111 1 31 1 22+7 0 0 1 6+21 0 0 0 0 0 1 1 5+20
11111110111111111111111111111111 1 31 1 23+6 0 0 1 5+22 0 0 0 0 0 1 1 4+21
11111101111111111111111111111111 1 31 1 24+5 0 0 1 4+23 0 0 0 0 0 1 1 3+22
11111011111111111111111111111111 1 31 1 25+4 0 0 1 3+24 0 0 0 0 0 1 1 2+23
11110111111111111111111111111111 1 31 1 26+3 0 0 1 2+25 0 0 0 0 0 1 1 1+24
11101111111111111111111111111111 1 31 1 27+2 0 0 1 1+26 0 0 0 0 0 1 1 25
11011111111111111111111111111111 1 31 1 28+1 0 0 1 27 0 0 0 0 0 1 0 26
10111111111111111111111111111111 1 31 1 29 0 0 1 28 0 0 0 0 0 1 0 27
01111111111111111111111111111111 1 31 0 30 0 0 1 29 0 0 0 0 0 1 0 28
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
32 31*32 31 29*30 0 0 32 27*28 0 0 0 0 0 31 29 25*26
+60 +114 +52+54+56
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
930 870 812 3759

n p 0 1 10 11 100 101 110 111 1000 1001 1010 1011 1100 1101 1110 1111
32 1 31
32 1
32 1
32 1
32 1
32 1
32 1
32 1
32 1
32 1
32 1
32 1
++++++++++++++++++++++++++++++++
12

11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111
11111111111111111111111111111111

dear google microsoft (cc intel)

regarding pattern recognition competition

first take a look at the diagonal of reason in another thread.
thirty two 32 bit strings of binary, with a zero diagonal, opposite cantors.
that is now called the square of reason.

use the square of reason to initially validate pattern recognition algorithms.
search for numbers (in binary) zero through fifteen inside the square of reason.
any correct competition entry must find 3759 occurrences of the number patterns.

after validation, entries just must answer with the number of occurrences
of the pattern in the stream or text.

for each (m, n) find (#memrefsm + #memrefsn)
winner of an mn point has least #memrefs

pure binary alphabet
small m, large n, purest random bits
m=0 n=0
m=0 n=1
m=1 n=2
m=2 n=4
m=3 n=8
m=4 n=16
m=5 n=32
m=8 n=64
m=11 n=128
m=16 n=256
m=23 n=512
m=32 n=1,000
m=1,000 n=1,000,000
m=10,000 n=1,000,000,000

m=8 n=1,000,000,000
m=16 n=1,000,000,000
m=32 n=1,000,000,000
m=64 n=1,000,000,000
m=128 n=1,000,000,000
m=256 n=1,000,000,000
m=512 n=1,000,000,000
m=4 n=1,000,000,000
m=2 n=1,000,000,000
m=1 n=1,000,000,000

for the competition count (number of pattrn references, number of text references)

since base three is most compact information storage
copy competition on alphabet size three (cc intel)
sincerely daniel

ps any point at all in point by point locations
of the solutions at this time (i prefer future work)

points of contention are these
purest of random bits
speed of random bits
speed of algorithm
speed of microsoft os
speed of google os
speed of daniels os
memory profiling
parallelism red herring for both of them not me
i already know how to do perfed parallelism and i decline to use it

strong color theorem: most regular graphs of degree d squared are d + one colorable.