Compressor: Squashes the dynamic range (difference between loudest and quietest parts of the music) - so that instead of, say 30db difference between the loudest sound (usually the vocal or lead instrument) and the quietest sound (usually FX and atmospherics) it's down to only 10dB (or less). This means that the loudest sound is still as loud as it was, but the quietest sound is now louder by comparison.
Think about how TV ads sound louder than the program, that's because they are heavily compressed so that the average overall volume is higher, though the loudest noise in the ad is no louder than the loudest noise in the program, unless it's Coro Street or a Miss Marple special
Limiter: Two types, soft-knee and hard-knee. A limiter stops the output volume passing through it from going above a pre-set threshold, if it does the sound is compressed, but in this case the loudest sound is stopped and everything else is coming up to meet it. Soft-knee applies the brakes gradually before the volume hits the treshold (best for music), hard-knee pimp-slaps the sound when it hits the threshold - often used for spoken word applications to stomp on mic pops caused by over-enthusiastic delivery of fricatives (look it up).
Gate: Most obvious example is the microphone dipping available on some mixers - if signal above a certain threshold comes down the mic the volume on the music is dropped to allow for clarity/audibility. The microphone signal closes, if only partially, the gate allowing the audio program through. Gates are also used in dance music to make the kick drum sound louder - each time the kick drum hits the volume on the rest of the music program drops slightly to accentuate the kick drum (Listen to Eric Prydz 'Call on Me' for an obvious use of this).
Feedback Destroyer: This a cross between an EQ and a filter. Feedback has a signature frequency which varies depending on the room, mostly, and the distance between the microphone and speakers - when feedback occurs output on the signature frequency increases out of step with the rest of the audio program a Feedback Destoyer establishes which frequency this is and using a notch-filter (a filter with a very narrow 'Q' or bandwidth) turns down the volume on just the effected frequency. leaving the rest of the audio untouched and with minimal effect on the overall sound.
I hope the descriptions make it clear when to use which.