In order to understand the lighting quality effects on the light-adaptive growth of cyanobacteria which does not express phycoerythrin, the thermophilic cyanobacteria PCC6715 in logarithmic growth phase was cultured by using white light (control), red light, and blue light with photosynthetically effective photon quantum density (PAR) of 100 μmol/(m²·s). Compared with the growth characteristics of PCC6715 under white light conditions, the changes in the measured and analyzed values of PCC6715 under other light qualities were as follows: the dry weight of PCC6715 significantly increased under blue light and significantly reduced under red light; the red and blue light have different effects on the start time and duration of the photosynthetic pigment synthesis of PCC6715; the maximum light energy conversion efficiency (F_v/F_m) of PCC6715 significantly increased under blue light and decreased significantly under red light; after 6 days of culture, the F_v/F_m of PCC6715 under red and blue light conditions tends to stable. In conclusion, the blue spectrum lighting promotes the growth of PCC6715, while red light inhibits the growth of PCC6715. At the same time, PCC6715 produces light-adaptive growth under both red and blue light, the reduction of phycocyanin content caused light-adaptive growth of PCC6715 in red light, and the light-adaptive growth of PCC6715 in blue light was due to the increase of phycocyanin content.

A piecewise convolutional neural network with gating mechanism is proposed, which would automatically filter positive correlation features at word-level. Moreover, the idea of soft-label is introduced to the gating mechanism to weaken the impact of hard labels on noise filtering. Combined with sentence-level noise filtering, the overall performance of the model is improved. The experimental results on the public dataset show that the proposed model has a significant improvement compared to the sentence-level noise filtering methods.