AI demonstrations
Associative memory
on a Commodore 64
John Walker’s 1987 BASIC program, recorded learning and recalling letter patterns.
What the program does
Walker’s program stores patterns in a weight matrix. During recall it repeatedly transforms the presented pattern until the result stops changing. The pattern converges during this recall process.
This early associative-memory demonstration stores and recalls simple patterns. Walker’s article includes the BASIC listing and an explanation of its limited capacity.
Reading the recording
- The demonstration teaches the network the letter patterns R and H.
- An unfamiliar A is presented. The recalled output is H.
- Noisy R and degraded H patterns are presented. The network recalls the corresponding learned patterns.
The two panes show the input and recalled patterns. The results illustrate how this particular demonstration behaves.
About this recording
Filmed from the collection’s Commodore 64 demonstration. The existing media filename refers to 1982; the program itself is dated September 4, 1987 in Walker’s publication.
Sources & records
- John Walker, “Neural Network on a Commodore 64,” September 4, 1987: Original article, method, sample output, and source listing.