Summary of Detecting Out-of-distribution Text Using Topological Features Of Transformer-based Language Models, by Andres Pollano et al.
Detecting out-of-distribution text using topological features of transformer-based language models
by Andres Pollano, Anupam Chaudhuri, Anj Simmons
First submitted to arxiv on: 22 Nov 2023
Categories
- Main: Computation and Language (cs.CL)
- Secondary: Machine Learning (cs.LG); Algebraic Topology (math.AT)
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 methodology utilizes topological features of self-attention maps from transformer-based language models to detect out-of-distribution (OOD) inputs. This approach is applicable to any transformer-based language model with multihead self-attention. The evaluation focuses on BERT, comparing it to a traditional OOD method using CLS embeddings. Results show that the proposed approach outperforms CLS embeddings in distinguishing in-distribution samples from far-OOD samples, but struggles with near or same-domain datasets. |
| Low | GrooveSquid.com (original content) | Low Difficulty Summary A team of researchers has developed a new way to keep machine learning systems safe from unexpected inputs that could cause problems. They used special features from language models like BERT to detect when text is unusual or outside the normal range. This method works well for very different texts, but has some trouble with texts that are similar but not exactly the same as the ones it was trained on. |
Keywords
* Artificial intelligence * Bert * Language model * Machine learning * Self attention * Transformer
