Arithmetic Coding is SO easy to understand. It's better than Huffman Coding. If
you only have 3 vocab words then you assign them bit codes as so a=00 b=01
c=10. If we store abcba, it costs 10 bits. Bit code 11 isn't used, we can do
better. So all you do is understand all combination 3*3*3*3*3=243 and so it
always is 7 bits (1.4 bits per character). In text, abcba costs 40 bits (8bpc).
To implement AC, you get the number by stepping through each vocab token
[a]bcba a[b]cba jumping into a new range as go ex. 243 divided by 3 = pick 1-81
or 82-162 or 163-243, repeat. This number ex. 129 can easily be converted
to/from binary bits.
------------------------------------------
Artificial General Intelligence List: AGI
Permalink:
https://agi.topicbox.com/groups/agi/T56d8c6758403c8e0-M3f76e1f3b66fb0540b8e37be
Delivery options: https://agi.topicbox.com/groups/agi/subscription
