| Title: | DEC RdbExpert for OpenVMS |
| Notice: | RdbExpert V2.1A is now available - See note 2.4 |
| Moderator: | OOTOOL::CRAIG |
| Created: | Fri Jul 10 1992 |
| Last Modified: | Tue Jun 03 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 755 |
| Total number of notes: | 2761 |
Has anyone ever looked at making RdbExpert work multiple passes
of instance data to check the deltas for progress ? Couldn't
we take a subset of the rules, such as number of buffers, and
let it automatically tune them ? If you remembered the results
wouldn't the product be able to rudimentarily "learn" ? Couldn't
such learning actually generate rules out of a large scale beta
test with real data ? There is a lot of potential for this in
Oracle7/8 as well where there is more to frob.
Regards,
rcs
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 750.1 | OOTOOL::GUAY | Oracle Expert, 1-2499 | Tue May 13 1997 09:15 | 14 | |
Exactly... In order to do this, we need to begin tracking changes to the instance parameters over time. Some of the capabilities that you describe will be introduced in 1.4 as the new "AutoTune" capability. Subsequent releases will move more in the direction that you outline. Generating results from cases (Case based reasoning), or from exemplars (Learning from examples), are both proven techniques for machine learning. Been brushing up on your AI Richard? :) | |||||
| 750.2 | Daydreams | NOMAHS::SECRIST | Rdb WWS; [email protected] | Tue May 13 1997 18:30 | 11 |
Thanks !
; Been brushing up on your AI Richard?
No, just using the products just before mowing my lawn, so I had
some extra CPU cycles...
Regards,
rcs
| |||||