paper-with-me

Papers

Towards Interpretable Natural Language Understanding with Explanations as Latent Variables

2020-10-24 · NeurIPS 2020 12 · Wangchunshu Zhou, Jinyi Hu, HANLIN ZHANG, Xiaodan Liang, Maosong Sun, Chenyan Xiong, Jian Tang

Recently generating natural language explanations has shown very promising results in not only offering interpretable explanations but also providing additional information and supervision for prediction. However, existing approaches usually require a large set of human annotated explanations for training while collecting a large set of explanations is not only time consuming but also expensive. In this paper, we develop a general framework for interpretable natural language understanding that requires only a small set of human annotated explanations for training. Our framework treats natural language explanations as latent variables that model the underlying reasoning process of a neural model. We develop a variational EM framework for optimization where an explanation generation module and an explanation-augmented prediction module are alternatively optimized and mutually enhance each other. Moreover, we further propose an explanation-based self-training method under this framework for semi-supervised learning. It alternates between assigning pseudo-labels to unlabeled data and generating new explanations to iteratively improve each other. Experiments on two natural language understanding tasks demonstrate that our framework can not only make effective predictions in both supervised and semi-supervised settings, but also generate good natural language explanation.

📄 PDF Abstract BibTeX arXiv:2011.05268

Code (1)

JamesHujy/ELV 공식 구현 pytorch

Tasks

Explanation GenerationNatural Language Understanding

Similar Papers 제목 키워드 기반

Latent Space Interpretation for Stylistic Analysis and Explainable Authorship Attribution

2024-09-11 · Milad Alshomary, Narutatsu Ri, Marianna Apidianaki, Ajay Patel 외

Recent state-of-the-art authorship attribution methods learn authorship representations of texts in a latent, non-interpretable space, hindering their usability in real-world applications. Our work proposes a novel appro…

Authorship Attribution

Automatically Interpreting Millions of Features in Large Language Models

2024-10-17 · Gonçalo Paulo, Alex Mallen, Caden Juang, Nora Belrose

While the activations of neurons in deep neural networks usually do not have a simple human-understandable interpretation, sparse autoencoders (SAEs) can be used to transform these activations into a higher-dimensional l…

Semantic SimilaritySemantic Textual Similarity

From latent factors to language: a user study on LLM-generated explanations for an inherently interpretable matrix-based recommender system

2025-09-23 · Maxime Manderlier, Fabian Lecron, Olivier Vu Thanh, Nicolas Gillis arxiv

We investigate whether large language models (LLMs) can generate effective, user-facing explanations from a mathematically interpretable recommendation model. The model is based on constrained matrix factorization, where…

Evaluating SAE interpretability without explanations

2025-07-11 · Gonçalo Paulo, Nora Belrose arxiv

Sparse autoencoders (SAEs) and transcoders have become important tools for machine learning interpretability. However, measuring how interpretable they are remains challenging, with weak consensus about which benchmarks …

Explanation Generation

Extractive Explanations for Interpretable Text Ranking

2021-06-23 · Jurek Leonhardt, Koustav Rudra, Avishek Anand

Neural document ranking models perform impressively well due to superior language understanding gained from pre-training tasks. However, due to their complexity and large number of parameters, these (typically transforme…

Document RankingRetrievalSentence