Summary of Scanner: Knowledge-enhanced Approach For Robust Multi-modal Named Entity Recognition Of Unseen Entities, by Hyunjong Ok et al.
SCANNER: Knowledge-Enhanced Approach for Robust Multi-modal Named Entity Recognition of Unseen Entities
by Hyunjong Ok, Taeho Kil, Sukmin Seo, Jaeho Lee
First submitted to arxiv on: 2 Apr 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Artificial Intelligence (cs.AI)
GrooveSquid.com Paper Summaries
GrooveSquid.com’s goal is to make artificial intelligence research accessible by summarizing AI papers in simpler terms. Each summary below covers the same AI paper, written at different levels of difficulty. The medium difficulty and low difficulty versions are original summaries written by GrooveSquid.com, while the high difficulty version is the paper’s original abstract. Feel free to learn from the version that suits you best!
| Summary difficulty | Written by | Summary |
|---|---|---|
| High | Paper authors | High Difficulty Summary Read the original abstract here |
| Medium | GrooveSquid.com (original content) | Medium Difficulty Summary The paper proposes SCANNER, a two-stage model for named entity recognition (NER) that can handle multiple variants, including multi-modal NER and grounded MNER. The model extracts entity candidates in the first stage and uses them as queries to retrieve knowledge from various sources. This approach allows it to generalize to unseen entities and address noisy annotations in training data. Additionally, the paper introduces a novel self-distillation method that enhances the robustness and accuracy of the model. SCANNER demonstrates competitive performance on NER benchmarks and surpasses existing methods on multi-modal and grounded MNER benchmarks. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary The paper is about a new way to find named entities in text and images. It’s like a detective trying to figure out who or what is being talked about. The new method, called SCANNER, can handle different types of information, like words, pictures, and sounds. It does this by looking at possible entity candidates and then using that information to learn more about the entities. This helps it find entities it hasn’t seen before, even if the training data has mistakes. The new approach works well on tests and is better than other methods for certain types of recognition. |
Keywords
» Artificial intelligence » Distillation » Multi modal » Named entity recognition » Ner
