Decision-directed learning

Robert Lucky's own term (from his 1966 BSTJ follow-up paper) for his adaptive equalizer's training mode, in which the receiver's own tentative decisions about already-received symbols are fed back as the error reference for adjusting the filter's tap gains after the known-training-sequence period ends — distinct in mechanism from the LMS algorithm's continuous use of a known reference signal. Surfaced as an entity candidate in the 2026-07-22 capture hunting for Lucky's own account of Widrow-Hoff awareness; not yet independently read from a primary source. Watching for recurrence before promotion to a full hub.

References

• claim-lucky-1965-adaptive-equalizer-steepest-descent-transversal-filter · entity-robert-w-lucky
• Capture: 2026-07-22-does-robert-luckys-own-oral-history-or-writing