Zero-Shot Learning with Knowledge Enhanced Visual Semantic Embeddings
This work provides consistent gains in zero-shot learning performance for researchers and practitioners working with limited labeled data by incorporating common-sense knowledge.
This paper tackles zero-shot learning by integrating common-sense knowledge into deep neural networks through a novel neuro-symbolic loss function. The proposed method, Common-Sense based Neuro-Symbolic Loss (CSNL), regularizes visual-semantic embeddings to adhere to hypernym and attribute relationships, achieving improvements of 11.5% on AWA2, 5.5% on CUB, and 11.6% on Kinetics over prior state-of-the-art.
We improve zero-shot learning (ZSL) by incorporating common-sense knowledge in DNNs. We propose Common-Sense based Neuro-Symbolic Loss (CSNL) that formulates prior knowledge as novel neuro-symbolic loss functions that regularize visual-semantic embedding. CSNL forces visual features in the VSE to obey common-sense rules relating to hypernyms and attributes. We introduce two key novelties for improved learning: (1) enforcement of rules for a group instead of a single concept to take into account class-wise relationships, and (2) confidence margins inside logical operators that enable implicit curriculum learning and prevent premature overfitting. We evaluate the advantages of incorporating each knowledge source and show consistent gains over prior state-of-art methods in both conventional and generalized ZSL e.g. 11.5%, 5.5%, and 11.6% improvements on AWA2, CUB, and Kinetics respectively.