Abstract
A general adaptive pattern recognition and learning software system, suitable for a broad range of applications, has been developed. This system integrates state-of-the-art methods in pattern recognition and learning in a software structure uniquely suited to the special characteristics of ballistic missile defense classication problems. The capability of this software was compared experimentally with that of an existing nonadaptive pattern recognition program. The experiments indicate that the adaptive software has advantages over nonadaptive software in adjusting to actual data characteristics, recognizing new classes of data, and learning rapidly, even with imperfect feedback.
| Original language | English |
|---|---|
| Pages | 785-790 |
| Number of pages | 6 |
| State | Published - 1977 |
| Event | COMPSAC '77, IEEE Comput Soc Int Comput Software & Appl Conf, 1st, Proc - Chicago, IL, USA Duration: 8 Nov 1977 → 11 Nov 1977 |
Conference
| Conference | COMPSAC '77, IEEE Comput Soc Int Comput Software & Appl Conf, 1st, Proc |
|---|---|
| City | Chicago, IL, USA |
| Period | 8/11/77 → 11/11/77 |