NeurIPS 2023 Papers

In this paper, we propose to approximate the softmax output, which is the key product of the attention mechanism, to reduce its activation memory usage when training attention-based networks (aka Transformers). During the forward pass of the network, the proposed softmax output approximation method stores only a small fraction of the entire softmax output required for back-propagation and evicts the rest of the softmax output from memory. Then, during the backward pass, the evicted softmax activation output is approximated to compose the gradient to perform back-propagation for model training. Considering most attention-based models heavily rely on the softmax-based attention module that usually takes one of the biggest portions of the network, approximating the softmax activation output can be a simple yet effective way to decrease the training memory requirement of many attention-based networks. The experiment with various attention-based models and relevant tasks, i.e., machine translation, text classification, and sentiment analysis, shows that it curtails the activation memory usage of the softmax-based attention module by up to 84% (6.2× less memory) in model training while achieving comparable or better performance, e.g., up to 5.4% higher classification accuracy.

The reliance of text classifiers on spurious correlations can lead to poor generalization at deployment, raising concerns about their use in safety-critical domains such as healthcare. In this work, we propose to use counterfactual data augmentation, guided by knowledge of the causal structure of the data, to simulate interventions on spurious features and to learn more robust text classifiers. We show that this strategy is appropriate in prediction problems where the label is spuriously correlated with an attribute. Under the assumptions of such problems, we discuss the favorable sample complexity of counterfactual data augmentation, compared to importance re-weighting. Pragmatically, we match examples using auxiliary data, based on diff-in-diff methodology, and use a large language model (LLM) to represent a conditional probability of text. Through extensive experimentation on learning caregiver-invariant predictors of clinical diagnoses from medical narratives and on semi-synthetic data, we demonstrate that our method for simulating interventions improves out-of-distribution (OOD) accuracy compared to baseline invariant learning algorithms.

As societal awareness of climate change grows, corporate climate policy engagements are attracting attention. We propose a dataset to estimate corporate climate policy engagement from various PDF-formatted documents. Our dataset comes from LobbyMap (a platform operated by global think tank InfluenceMap) that provides engagement categories and stances on the documents. To convert the LobbyMap data into the structured dataset, we developed a pipeline using text extraction and OCR. Our contributions are: (i) Building an NLP dataset including 10K documents on corporate climate policy engagement. (ii) Analyzing the properties and challenges of the dataset. (iii) Providing experiments for the dataset using pre-trained language models. The results show that while Longformer outperforms baselines and other pre-trained models, there is still room for significant improvement. We hope our work begins to bridge research on NLP and climate change.

Backdoor attacks have emerged as a prominent threat to natural language processing (NLP) models, where the presence of specific triggers in the input can lead poisoned models to misclassify these inputs to predetermined target classes. Current detection mechanisms are limited by their inability to address more covert backdoor strategies, such as style-based attacks. In this work, we propose an innovative test-time poisoned sample detection framework that hinges on the interpretability of model predictions, grounded in the semantic meaning of inputs. We contend that triggers (e.g., infrequent words) are not supposed to fundamentally alter the underlying semantic meanings of poisoned samples as they want to stay stealthy. Based on this observation, we hypothesize that while the model's predictions for paraphrased clean samples should remain stable, predictions for poisoned samples should revert to their true labels upon the mutations applied to triggers during the paraphrasing process. We employ ChatGPT, a state-of-the-art large language model, as our paraphraser and formulate the trigger-removal task as a prompt engineering problem. We adopt fuzzing, a technique commonly used for unearthing software vulnerabilities, to discover optimal paraphrase prompts that can effectively eliminate triggers while concurrently maintaining input semantics. Experiments on 4 types of backdoor attacks, including the subtle style backdoors, and 4 distinct datasets demonstrate that our approach surpasses baseline methods, including STRIP, RAP, and ONION, in precision and recall.