Reasoning over Vision and Language: Exploring the Benefits of Supplemental Knowledge

Abstract

The limits of applicability of vision-and-language models are defined by the coverage of their training data. Tasks like vision question answering (VQA) often require commonsense and factual information beyond what can be learned from task-specific datasets. This paper investigates the injection of knowledge from general-purpose knowledge bases (KBs) into vision-and-language transformers. We use an auxiliary training objective that encourages the learned representations to align with graph embeddings of matching entities in a KB. We empirically study the relevance of various KBs to multiple tasks and benchmarks. The technique brings clear benefits to knowledge-demanding question answering tasks (OK-VQA, FVQA) by capturing semantic and relational knowledge absent from existing models. More surprisingly, the technique also benefits visual reasoning tasks (NLVR2, SNLI-VE). We perform probing experiments and show that the injection of additional knowledge regularizes the space of embeddings, which improves the representation of lexical and semantic similarities. The technique is model-agnostic and can expand the applicability of any vision-and-language transformer with minimal computational overhead.

References in corpus (22)

• Google's Neural Machine Translation System: Bridging the Gap between Human and Machine Translation
• Microsoft COCO Captions: Data Collection and Evaluation Server
• VisualBERT: A Simple and Performant Baseline for Vision and Language
• ConceptNet 5.5: An Open Multilingual Graph of General Knowledge
• VL-BERT: Pre-training of Generic Visual-Linguistic Representations
• Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challenge
• A large annotated corpus for learning natural language inference
• SentEval: An Evaluation Toolkit for Universal Sentence Representations
• CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge
• LXMERT: Learning Cross-Modality Encoder Representations from Transformers
• UNITER: UNiversal Image-TExt Representation Learning
• Visual Entailment: A Novel Task for Fine-Grained Image Understanding
• ERNIE: Enhanced Language Representation with Informative Entities
• Unicoder-VL: A Universal Encoder for Vision and Language by Cross-modal Pre-training
• Explicit Knowledge-based Reasoning for Visual Question Answering
• Bilinear Attention Networks
• Align, Mask and Select: A Simple Method for Incorporating Commonsense Knowledge into Language Representation Models
• Recent Advances in Natural Language Inference: A Survey of Benchmarks, Resources, and Approaches
• Zero-Shot Visual Question Answering
• Are we pretraining it right? Digging deeper into visio-linguistic pretraining
• Knowledge Enhanced Contextual Word Representations
• Bridging the Knowledge Gap: Enhancing Question Answering with World and Domain Knowledge