Skip to search formSkip to main contentSkip to account menu

VisualBERT: A Simple and Performant Baseline for Vision and Language

Analysis demonstrates that VisualBERT can ground elements of language to image regions without any explicit supervision and is even sensitive to syntactic relationships, tracking, for example, associations between verbs and image regions corresponding to their arguments.
View PDF on arXiv (opens in a new tab)

Grounded Language-Image Pre-training

A grounded language-image pretraining model for learning object-level, language-aware, and semantic-rich visual representations that unifies object detection and phrase grounding for pre-training and can leverage massive image-text pairs by generating grounding boxes in a self-training fashion.
View on IEEE (opens in a new tab)

How Much Can CLIP Benefit Vision-and-Language Tasks?

It is shown that CLIP significantly outperforms widely-used visual encoders trained with in-domain annotated data, such as BottomUp-TopDown, and also establishes new state-of-the-art results on Visual Question Answering, Visual Entailment, and V&L Navigation tasks.
View PDF on arXiv (opens in a new tab)

What Does BERT with Vision Look At?

It is demonstrated that certain attention heads of a visually grounded language model actively ground elements of language to image regions, performing the task known as entity grounding.
View on ACL (opens in a new tab)

On the Paradox of Learning to Reason from Data

This study pro-vides an explanation for this paradox: instead of learning to emulate the correct reasoning function, BERT has in fact learned statistical features that inherently exist in logical reasoning problems.
View PDF on arXiv (opens in a new tab)

Point Precisely: Towards Ensuring the Precision of Data in Generated Texts Using Delayed Copy Mechanism

This paper proposes a two-stage approach with a delayed copy mechanism to improve the precision of data records in the generated texts and verifies the efficacy of the proposed approach to generate better templates and copy data records more precisely.
View on ACL (opens in a new tab)

Efficient Contextual Representation Learning With Continuous Outputs

This work revisits the design of the output layer and considers directly predicting the pre-trained embedding of the target word for a given context and achieves a 4-fold speedup and eliminates 80% trainable parameters while achieving competitive performance on downstream tasks.
View on MIT Press (opens in a new tab)

Weakly-supervised VisualBERT: Pre-training without Parallel Images and Captions

This work proposes Weakly-supervised VisualBERT with the key idea of conducting "mask-and-predict" pre-training on language-only and image-only corpora, and introduces the object tags detected by an object recognition model as anchor points to bridge two modalities.
View PDF on arXiv (opens in a new tab)

Efficient Contextual Representation Learning Without Softmax Layer

This work redesigns the learning objective and proposes an efficient framework for training contextual representation models that bypasses the softmax layer by performing language modeling with dimension reduction, and allows the models to leverage pre-trained word embeddings.
View PDF on arXiv (opens in a new tab)

Leveraging Diverse Lexical Chains to Construct Essays for Chinese College Entrance Examination

A sentence extraction framework based on diversified lexical chains to capture coherence and richness is explored and reveals the importance of information richness in essay writing.
View on ACL (opens in a new tab)
By clicking accept or continuing to use the site, you agree to the terms outlined in our Privacy Policy (opens in a new tab), Terms of Service (opens in a new tab), and Dataset License (opens in a new tab)