A typical choice is the Rectified Linear Unit (ReLU) activation function. To address over-fitting in the existing ReLU-based NMD model (ReLU-NMD), we propose a Tikhonov regularized ReLU-NMD model, ...

In previous studies, textual information and graph structural information are utilized independently, without an effective method for fusing these two types of information. In this paper, we propose a ...

Unlike traditional databases, knowledge graphs organize information as nodes and edges, making them better for AI systems that reason & infer.