Abstract: Drug target affinity prediction plays an important role in drug development. However, most existing prediction methods neglect the two-dimensional structure information of drug molecules and lack deep representation fusion learning. To address this issue, a drug target affinity prediction model (GBN_DTA) based on graph convolutional network and bilinear attention network is proposed. This model first encodes the drug molecule graph using a multi-layer graph convolutional neural network, and simultaneously encodes the target sequence using 1D-CNN and BiLSTM. Then, a bilinear attention network is used to fuse the encoded drug and target features to obtain the affinity prediction score. Experimental results show that the performance of this model on the DAVIS and KIBA datasets is superior to six other mainstream methods, effectively improving the prediction accuracy.