Summary of Incremental Sequence Labeling: a Tale Of Two Shifts, by Shengjie Qiu et al.
Incremental Sequence Labeling: A Tale of Two Shifts
by Shengjie Qiu, Junhao Zheng, Zhen Liu, Yicheng Luo, Qianli Ma
First submitted to arxiv on: 16 Feb 2024
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG)
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 proposed framework, Incremental Sequential Labeling without Semantic Shifts (IS3), tackles the incremental sequence labeling task by addressing two significant semantic shifts: E2O and O2E. Previous research has primarily focused on mitigating E2O, neglecting O2E, which leads to model bias towards new entities during learning. IS3 aims to alleviate this bias through debiased loss and optimization levels. Additionally, knowledge distillation is employed to maintain the model’s discriminative ability for old entities. Experimental results on three datasets demonstrate superior performance compared to the previous state-of-the-art method. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary Incremental sequence labeling helps machines learn new classes over time while remembering previous ones. Researchers found two big problems: E2O (old entity mislabeled as non-entity) and O2E (non-entity or old entity labeled as new). Most research focused on fixing E2O, ignoring O2E. This caused models to be biased towards labeling new data as belonging to the new class. To solve this, scientists created a new framework called IS3. IS3 tries to fix both E2O and O2E problems by using special techniques like knowledge distillation and debiased loss. |
Keywords
* Artificial intelligence * Knowledge distillation * Optimization
