Learning Visually-Grounded Semantics from Contrastive Adversarial Samples
This work addresses the limitation of visual-semantic embeddings for AI systems that rely on multimodal understanding, though it is incremental as it builds on existing frameworks.
The authors tackled the problem of grounding textual semantics in visual concepts by augmenting image captioning datasets with contrastive adversarial samples, resulting in noticeable improvements on downstream tasks and enhanced defense against adversarial attacks.
We study the problem of grounding distributional representations of texts on the visual domain, namely visual-semantic embeddings (VSE for short). Begin with an insightful adversarial attack on VSE embeddings, we show the limitation of current frameworks and image-text datasets (e.g., MS-COCO) both quantitatively and qualitatively. The large gap between the number of possible constitutions of real-world semantics and the size of parallel data, to a large extent, restricts the model to establish the link between textual semantics and visual concepts. We alleviate this problem by augmenting the MS-COCO image captioning datasets with textual contrastive adversarial samples. These samples are synthesized using linguistic rules and the WordNet knowledge base. The construction procedure is both syntax- and semantics-aware. The samples enforce the model to ground learned embeddings to concrete concepts within the image. This simple but powerful technique brings a noticeable improvement over the baselines on a diverse set of downstream tasks, in addition to defending known-type adversarial attacks. We release the codes at https://github.com/ExplorerFreda/VSE-C.