An official website of the United States government
Existing backdoor defense methods are only effective for limited trigger types. To defend different trigger types at once, we start from the class-irrelevant nature of the poisoning process and propose a novel weakly supervised backdoor defense framework WeDef. Recent advances in weak supervision make it possible to train a reasonably accurate text classifier using only a small number of user-provided, class-indicative seed words. Such seed words shall be considered independent of the triggers. Therefore, a weakly supervised text classifier trained by only the poisoned documents without their labels will likely have no backdoor. Inspired by this observation, in WeDef, we define the reliability of samples based on whether the predictions of the weak classifier agree with their labels in the poisoned training set. We further improve the results through a two-phase sanitization: (1) iteratively refine the weak classifier based on the reliable samples and (2) train a binary poison classifier by distinguishing the most unreliable samples from the most reliable samples. Finally, we train the sanitized model on the samples that the poison classifier predicts as benign. Extensive experiments show that WeDef is effective against popular trigger-based attacks (e.g., words, sentences, and paraphrases), outperforming existing defense methods.
more »
« less
“Misc”-Aware Weakly Supervised Aspect Classification
Aspect classification, identifying aspects of text segments, facilitates numerous applications, such as sentiment analysis and review summarization. To alleviate the extensive human effort required by existing aspect classification methods, in this paper, we focus on a weakly supervised setting—the model input only contains domainspecific raw texts and a few seed words per pre-defined aspect. We identify a unique challenge here as to how to classify texts without any pre-defined aspects. The existence of this kind of “misc” aspect text segments is very common in review corpora. It is difficult, even for domain experts, to nominate seed words for the “misc” aspect, which makes existing seed-driven text classification methods not applicable. Therefore, we propose to jointly model pre-defined aspects and the “misc” aspect through a novel framework, ARYA. It enables mutual enhancements between pre-defined aspects and the “misc” aspect via iterative classifier training and seed set updating. Specifically, it trains a classifier for pre-defined aspects and then leverages it to induce the supervision for the “misc” aspect. The prediction results of the “misc” aspect are later utilized to further filter the seed word selections for pre-defined aspects. Experiments in three domains demonstrate the superior performance of our proposed framework, as well as the necessity and importance of properly modeling the “misc” aspect
more »
« less
- Award ID(s):
- 2040727
- PAR ID:
- 10250423
- Date Published:
- Journal Name:
- Proceedings of the 2021 SIAM International Conference on Data Mining (SDM)
- Format(s):
- Medium: X
- Sponsoring Org:
- National Science Foundation
More Like this
-
-
We design a generic framework for learning a robust text classification model that achieves high accuracy under different selection budgets (a.k.a selection rates) at test-time. We take a different approach from existing methods and learn to dynamically filter a large fraction of unimportant words by a low-complexity selector such that any high-complexity state-of-art classifier only needs to process a small fraction of text, relevant for the target task. To this end, we propose a data aggregation method to train the classifier, allowing it to achieve competitive performance on fractured sentences. On four benchmark text classification tasks, we demonstrate that the framework gains consistent speedup with little degradation in accuracy on various selection budgets.more » « less
-
null (Ed.)Abstract Sentiment, judgments and expressed positions are crucial concepts across international relations and the social sciences more generally. Yet, contemporary quantitative research has conventionally avoided the most direct and nuanced source of this information: political and social texts. In contrast, qualitative research has long relied on the patterns in texts to understand detailed trends in public opinion, social issues, the terms of international alliances, and the positions of politicians. Yet, qualitative human reading does not scale to the accelerating mass of digital information available currently. Researchers are in need of automated tools that can extract meaningful opinions and judgments from texts. Thus, there is an emerging opportunity to marry the model-based, inferential focus of quantitative methodology, as exemplified by ideal point models, with high resolution, qualitative interpretations of language and positions. We suggest that using alternatives to simple bag of words (BOW) representations and re-focusing on aspect-sentiment representations of text will aid researchers in systematically extracting people’s judgments and what is being judged at scale. The experimental results below show that our approach which automates the extraction of aspect and sentiment MWE pairs, outperforms BOW in classification tasks, while providing more interpretable parameters. By connecting expressed sentiment and the aspects being judged, PULSAR (Parsing Unstructured Language into Sentiment-Aspect Representations) also has deep implications for understanding the underlying dimensionality of issue positions and ideal points estimated with text. Our approach to parsing text into aspects-sentiment expressions recovers both expressive phrases (akin to categorical votes), as well as the aspects that are being judged (akin to bills). Thus, PULSAR or future systems like it, open up new avenues for the systematic analysis of high-dimensional opinions and judgments at scale within existing ideal point models.more » « less
-
null (Ed.)Recent advances in weakly supervised learn- ing enable training high-quality text classifiers by only providing a few user-provided seed words. Existing methods mainly use text data alone to generate pseudo-labels despite the fact that metadata information (e.g., author and timestamp) is widely available across various domains. Strong label indicators exist in the metadata and it has been long overlooked mainly due to the following challenges: (1) metadata is multi-typed, requiring systematic modeling of different types and their combinations, (2) metadata is noisy, some metadata entities (e.g., authors, venues) are more compelling label indicators than others. In this paper, we propose a novel framework, META, which goes beyond the existing paradigm and leverages metadata as an additional source of weak supervision. Specifically, we organize the text data and metadata together into a text-rich network and adopt network motifs to capture appropriate combinations of metadata. Based on seed words, we rank and filter motif instances to distill highly label-indicative ones as “seed motifs”, which provide additional weak supervision. Following a boot-strapping manner, we train the classifier and expand the seed words and seed motifs iteratively. Extensive experiments and case studies on real-world datasets demonstrate superior performance and significant advantages of leveraging metadata as weak supervision.more » « less
-
null (Ed.)Aspect-based sentiment analysis of review texts is of great value for understanding user feedback in a fine-grained manner. It has in general two sub-tasks: (i) extracting aspects from each review, and (ii) classifying aspect-based reviews by sentiment polarity. In this pa-per, we propose a weakly-supervised approach for aspect-based sentiment analysis, which uses only a few keywords describing each aspect/sentiment without using any labeled examples. Existing methods are either designed only for one of the sub-tasks, neglecting the benefit of coupling both, or are based on topic models that may contain overlapping concepts. We propose to first learn sentiment, aspectjoint topic embeddings in the word embedding space by imposing regularizations to encourage topic distinctiveness, and then use neural models to generalize the word-level discriminative information by pre-training the classifiers with embedding-based predictions and self-training them on unlabeled data. Our comprehensive performance analysis shows that our method generates quality joint topics and outperforms the baselines significantly (7.4%and 5.1% F1-score gain on average for aspect and sentiment classification respectively) on benchmark datasets.more » « less
- Free Publicly Accessible Full Text
-
Accepted Manuscript1.0
- Conference Paper:
- The DOI is not currently available.
