Discretized Integrated Gradients for Explaining Language Models (2021.emnlp-main)
Copied to clipboard
| Challenge: | Integrated Gradients (IG) is widely adopted due to its desirable explanation axioms and the ease of gradient computation. |
| Approach: | They propose an attribution-based explanation algorithm that uses averaging the model's output gradient interpolated along a straight-line path in the input data space. |
| Outcome: | The proposed method is compared with IG on multiple sentiment classification datasets. |
Similar Papers
Sequential Integrated Gradients: a simple but effective method for explaining language models (2023.findings-acl)
Copied to clipboard
| Challenge: | Existing explanation methods such as Integrated Gradients (IG) produce a path for each word of a sentence simultaneously, which can lead to sentences with no clear meaning or a significantly different meaning compared to the original one. |
| Approach: | They propose to use a sequenced integrated gradient method to fix every word in a sentence and move it along a straight path to the word of interest. |
| Outcome: | The proposed method improves on Integrated Gradients (IG) and DIG, but can produce sentences with different meanings than the original one. |
Integrated Directional Gradients: Feature Interaction Attribution for Neural NLP Models (2021.acl-long)
Copied to clipboard
| Challenge: | Existing methods for attribution of importance to features borrowed from cooperative game theory . success of Deep Neural Networks has led to their ability to learn from complex higher order interactions from raw features. |
| Approach: | They propose a method for attributing importance scores to groups of features . they propose axioms that any intuitive feature group attribution method should satisfy . |
| Outcome: | The proposed method captures the importance of features in a linguistic model using negations and conjunctions. |
Explaining Classes through Stable Word Attributions (2022.findings-acl)
Copied to clipboard
| Challenge: | Input saliency methods have become popular for explaining predictions of deep learning models, but there has been little work investigating methods for aggregating prediction-level explanations to the class level. |
| Approach: | They propose a method to aggregate prediction-level explanations to the class level using XLM-R and Integrated Gradients input attribution methods. |
| Outcome: | The proposed method extracts keyword lists of classes from text classification tasks and evaluates them on web register data. |
Latent Concept-based Explanation of NLP Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing attempts to explain deep learning models rely on input features, such as the words . however, such explanations are often less informative due to the discrete nature of words and lack of contextual verbosity. |
| Approach: | They propose a method that generates explanations for predictions based on latent concepts . they map the representations of salient input words into the training latent space . |
| Outcome: | The proposed method generates explanations for predictions based on latent concepts . it maps representations of salient input words into training latent space . |
TIGS: An Inference Algorithm for Text Infilling with Gradient Search (P19-1)
Copied to clipboard
| Challenge: | Text infilling is an under-explored challenge in the field of text generation. |
| Approach: | They propose an iterative inference algorithm based on gradient search that can be broadly applied to any sequence generative model for text infilling tasks. |
| Outcome: | The proposed method performs well on three different text infilling tasks with different mask ratios and mask strategies compared with five state-of-the-art methods. |
Multi-Level Explanations for Generative Language Models (2025.acl-long)
Copied to clipboard
Lucas Monteiro Paes, Dennis Wei, Hyo Jin Do, Hendrik Strobelt, Ronny Luss, Amit Dhurandhar, Manish Nagireddy, Karthikeyan Natesan Ramamurthy, Prasanna Sattigeri, Werner Geyer, Soumya Ghosh
| Challenge: | Large language models (LLMs) are being used for context-grounded tasks like summarizing meetings and answering doctors' questions. |
| Approach: | They propose a technique to provide explanations for context-grounded text generation by assigning scores to parts of the context to quantify their influence on the model output. |
| Outcome: | The proposed framework can provide more faithful explanations of generated output than available alternatives, including LLM self-explanations. |
Gradient-based Analysis of NLP Models is Manipulable (2020.findings-emnlp)
Copied to clipboard
| Challenge: | Recent work has shown that explanation techniques can be unstable and can be manipulated to hide the actual reasoning behind the predictions of NLP models. |
| Approach: | They propose to merge a BERT-based sentiment classifier with a Facade Model that overwhelms the gradients without affecting the predictions. |
| Outcome: | The proposed model overwhelms the gradients without affecting the predictions on a variety of NLP tasks, such as sentiment analysis, NLI, and QA. |
The Impact of Integration Step on Integrated Gradients (2024.eacl-srw)
Copied to clipboard
| Challenge: | Integrated gradients (IG) are a powerful tool for explaining the internal structure of a language model. |
| Approach: | They propose to customize the step count for each instance to minimize the error. |
| Outcome: | The optimal number of steps to maintain minimal error varies from instance to instance. |
Fine-grained Pluggable Gradient Ascent for Knowledge Unlearning in Language Models (2024.emnlp-main)
Copied to clipboard
| Challenge: | Existing studies on knowledge unlearning focus on computer vision but extend their exploration to other fields. |
| Approach: | They propose an adaptive objective that calculates gradients with fine-grained control specifically targeting sensitive tokens. |
| Outcome: | The proposed method improves the general ability of language models while achieving knowledge unlearning. |
Layered Insights: Generalizable Analysis of Human Authorial Style by Leveraging All Transformer Layers (2025.emnlp-main)
Copied to clipboard
| Challenge: | Existing approaches to authorship attribution model only learn from the output layer of pre-trained transformers, ignoring representations learned at other layers. |
| Approach: | They propose a model that leverages the various linguistic representations learned at different layers of pre-trained transformer-based models to model the authorship attribution task more effectively. |
| Outcome: | The proposed model performs better on out-of-domain and in-domain scenarios, while ignoring representations learned at other layers. |