We attempt to detect out-of-distribution (OOD) text samples though applying
Topological Data Analysis (TDA) to attention maps in transformer-based language
models. We evaluate our proposed TDA-based approach for out-of-distribution
detection on BERT, a transformer-based language model, and compare the to a
more traditional OOD approach based on BERT CLS embeddings. We found that our
TDA approach outperforms the CLS embedding approach at distinguishing
in-distribution data (politics and entertainment news articles from HuffPost)
from far out-of-domain samples (IMDB reviews), but its effectiveness
deteriorates with near out-of-domain (CNN/Dailymail) or same-domain (business
news articles from HuffPost) datasets.

通过在基于 transformer 的语言模型中的注意力图上应用拓扑数据分析（TDA）来检测超出分布（OOD）的文本样本。我们评估了我们提出的基于 TDA 的 OOD 检测方法在 BERT 上的效果，并与基于 BERT CLS 嵌入的传统 OOD 方法进行了比较。我们发现，我们的 TDA 方法在区分内部数据（HuffPost 的政治和娱乐新闻文章）与远离领域样本（IMDB 评论）方面优于 CLS 嵌入方法，但其效果在接近领域（CNN / Dailymail）或相同领域（HuffPost 的商业新闻文章）数据集上逐渐减弱。