We construct a catalogue of stellar masses and ages for 696680 red giant branch (RGB) stars, 180436 primary red clump (RC) stars, and 120907 secondary RC stars selected from the LAMOSTDR8. The RGBs, primary RCs, and secondary RCs are identified with the large frequency spacing ({DELTA}{nu}) and period spacing ({DELTA}P), estimated from the LAMOST spectra with spectral SNRs>10 by the neural network method supervised with the seismologic information from LAMOST-Kepler sample stars. The purity and completeness of both RGB and RC samples are better than 95% and 90%, respectively. The mass and age of RGBs and RCs are determined again with the neural network method by taking the LAMOST-Kepler giant stars as the training set. The typical uncertainties of stellar mass and age are, respectively, 10% and 30% for the RGB stellar sample. For RCs, the typical uncertainties of stellar mass and age are 9% and 24%, respectively. The RGB and RC stellar samples cover a large volume of the Milky Way (5<R<20kpc and |Z|<5 kpc), which are valuable data sets for various Galactic studies.