Tensor rank one differential graph preserving analysis for facial expression recognition

► A new dimensionality reduction algorithm is proposed based on tensor rank-one analysis and graph preserving. ► A novel marginal neighboring graph is devised to describe the pairwise inter-class boundaries. ► The algorithm is verified convergence during an iterative process. ► The algorithm consumes relatively less time than the vector representation and the tensor-to-tensor projection algorithms. ► The algorithm is effective for the facial expression recognition especially when followed by a neural network classifier.