What to do if a transistor unstable at an un-desired frequency

I'm designing an amplifier with an HEMT. At desired frequency, it looks good...

but the stable factor shows that it would be unstable at some undesired frequency... What should I do? Improve its stable factor? or ignore it?

First select a HEMT having its GBW optimal at the desired frequency, not higher. Next use low-pass filter(s) to limit the amplifier bandwidth to the desired one, not higher.
Most instabilities in amplifiers happen at higher than desired frequency.
Use a "clean" microstrip structure, check for possible higher modes and suppress these by absorbers. Most amplifiers need some tuning when tested for an optimum response.

Many low-noise and power amplifiers behave well when tested with sources and loads at close to 50 Ohms. When a source or load are mismatched, amplifiers tend to oscillate, often out-of-band. Isolators can be used with a caution, they can be mismatched at off-band, too.

And if the instability happens at the low frequency, what should I do then?

Low-frequency instability is rather an indication of a poor design. Use larger DC blocking capacitors, and use a high-pass filter for signal path.
Adjust the values of signal coupling capacitors, use inductors calculated to match output impedance.

Amplifier response should be designed to correspond to signal spectrum. Extending the response invites instability.

Input and output matching networks plays important role for the stability of the amplifier.
Using high-pass topology for the input network, and low-pass for the output network helps to avoid low frequency (or very high frequency) instabilities.

well Guys I have heard and also tested(in ADS), that adding a small resistor (2 ohm) in the signal path to the gate of the transistor resulted in unconditional stability.
what about that is it a good technique or not?

There are many methods that can be used to stabilize an amplifier. Take a look to this article about mosfets.

http://www.polyfet.com/MTT97.pdf