Training method and system of small classification model based on cross-modal migration knowledge data

• Main Authors: FU SIMING, LI LEI, WANG GUOHUA
• Format: Patent
• Language: Chinese; English
• Published: 14.05.2024
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; ELECTRIC DIGITAL DATA PROCESSING; PHYSICS

The invention relates to the technical field of model data training, in particular to a training method and system of a small classification model based on cross-modal migration knowledge data. The method comprises the following steps: obtaining structured data and unstructured data; performing knowledge graph construction on the structured data and the unstructured data to generate multi-dimensional heterogeneous information graph data; performing low-dimensional vector conversion on the multi-dimensional heterogeneous information atlas data to generate entity relationship embedded vector data; performing multi-modal fusion network structure design on the standard structured data and the standard unstructured data to generate a multi-modal adaptive fusion network structure; and performing multi-modal feature data fusion on the standard structured data and the standard unstructured data by using the multi-modal adaptive fusion network structure to generate multi-modal fusion feature adjustment data. According