The basic data for an airplane is a typical vector. So I simply feed the SOM with airplane data (several T/W and wing loading data and some other stuff), and let the artificial neurons do their job making some maps.
On the first picture, the map became devided in two distinct regions. "Real" fighters are on the left, while more fighter bomber planes are on the right. The Gripen and the Rafale are also somewhat isolated from the rest of the fighters. The F-35 is very similar to the F-18E. So, the F-35 is for all practical purposes (practical in a T/W ratio, wing loading sence :-) ; a stealthy F-18E. If this is good or bad, I don't know, but the F-35 is certainly no real fighter in a "classical" sense relative to the fighter bunch.
On the other picture I have "hard wired" a mystery plane in the middle and let the SOM work to see the similarities with the others. Clarly the mystery plane is very different from the rest, probably much better looking :-) But, it has to be built...
![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEjMqVd8mgL1ZvguNLw0D_2nPK1H6i1gExu10Tb_UtjZ1oLuJ08IkkeXgA4Swz3tmln_ruxM-ZOu3rNYwCPumh8kuPXqTZ0LRbW-OudRtDPhT1YFw8tyjUvApJ03p1CSFdExdlQzBATEX0bP/s320/fighters.jpg)
![](https://blogger.googleusercontent.com/img/b/R29vZ2xl/AVvXsEgPKY2lDalg7niprJqP8YKT8KdFjH93_jWFnrE22r_z2qp8MBglWUIO_MUjBS-ruOhiKciEE12zGNpuTDBdTegFfGw2W7qRg7dk6HLHMbPD25D0ga3a8rTEsegvFjbkbi1oYM0rVIH4z0Kf/s320/rv-4-fight2.jpg)