Interpretable Discriminative Text Representations via Agreement and Label Disentanglement

• URL: http://arxiv.org/abs/2605.20693v1
• Date: Wed, 20 May 2026 04:41:44 GMT
• Title: Interpretable Discriminative Text Representations via Agreement and Label Disentanglement
• Abstract summary: We propose an operational criterion for interpretable discriminative text representations.<n>Each coordinate should satisfy conceptual clarity, measured by chance of agreement.<n>We instantiate this criterion in LLM-assisted Feature Discovery.
• Score: 4.2788570168072395
• License: http://creativecommons.org/licenses/by-nc-sa/4.0/
• Abstract: Interpretable text representations should expose coordinates that are not only predictive, but also meaningful enough for independent auditors to apply. Existing discriminative representations often use anonymous embedding directions, while concept-bottleneck and LLM-assisted methods attach natural-language names to features without ensuring that those definitions are reproducible or distinct from the target label. We propose an operational criterion for interpretable discriminative text representations: each coordinate should satisfy conceptual clarity, measured by chance-adjusted agreement between independent annotators applying the feature definition, and label disentanglement, meaning the feature should not merely paraphrase the prediction target. We instantiate this criterion in LLM-assisted Feature Discovery (LFD), an iterative method that proposes lexical and semantic features from contrastive outcome-opposed text pairs, screens candidates using cross-LLM Cohen's $κ$, and selects features by residual held-out predictive gain. A stylized analysis connects the $κ$ screen to a per-feature annotation-noise bound, formalizing agreement as a reliability check. Across ten text-classification tasks spanning seven corpora, LFD matches the predictive performance of a strong text bottleneck baseline while producing substantially clearer and less label-entangled features. Human audits with 232 raters show that LFD features achieve higher human--human and human--LLM agreement than baseline concepts, and raters consistently judge them as less label-leaking. These results suggest that agreement-tested, label-disentangled coordinates provide a practical auditability standard for interpretable text classification.

Related papers

• Structuring Semantic Embeddings for Principle Evaluation: A Prototype-Guided Contrastive Learning Approach [58.15427952222424]
  General-purpose text embeddings are widely deployed for such tasks, but broad semantic similarity can place semantically similar yet task-distinct examples.<n>We introduce Prototype-Guided Contrastive Learning (PGCL), a prototype-guided geometric regularization module built on top of frozen text embeddings.
  arXiv  Detail & Related papers  (2026-08-15T13:19:25Z)
• Evaluating the Interpretability of Sparse Autoencoders with Concept Annotations [12.535445487099393]
  We present a human-grounded evaluation framework that quantifies alignment between SAE latents and human-annotated concepts.<n>We validate this matching through targeted attribute perturbations.<n>Our framework suggests that moderate dictionary sizes provide the best trade-off, yielding the most interpretable SAEs.
  arXiv  Detail & Related papers  (2026-06-23T15:39:29Z)
• A Finetuned SpeechLLM for Joint Multi-Granular L2 Assessment and Natural-Language Rationales [8.300738063140129]
  We propose a rubric-guided SpeechLLM for multi-aspect, multi-granular assessment.<n>The model jointly predicts ordinal labels at the sentence-level (accuracy, fluency, prosody), word/phoneme-level accuracy, and generates a natural-language rationale in the same response.
  arXiv  Detail & Related papers  (2026-06-08T13:27:40Z)
• Human Label Variation as Stable Signal: Learning Annotator-Specific Explanation Behavior via Cross-Annotator Preference Optimization [51.93456979139756]
  Free-text explanations extend human label variation beyond label disagreement.<n>We study whether large language models can learn and reproduce such annotator-specific label-explanation behavior.
  arXiv  Detail & Related papers  (2026-05-27T17:55:00Z)
• SCALE: Semantic- and Confidence-Aware Conditional Variational Autoencoder for Zero-shot Skeleton-based Action Recognition [4.853241666510524]
  Zero-shot skeleton-based action recognition (ZSAR) aims to recognize action classes without any training skeletons from those classes.<n>We propose a lightweight and deterministic Semantic- and Confidence-Aware Listwise Energy-based framework that formulates ZSAR as class-conditional energy ranking.
  arXiv  Detail & Related papers  (2026-04-02T16:12:42Z)
• A Straightforward Pipeline for Targeted Entailment and Contradiction Detection [0.15229257192293197]
  Key challenge is to identify which sentences act as premises or contradictions for a specific claim.<n>We introduce a method that combines the strengths of both approaches for a targeted analysis.<n>By filtering NLI-identified relationships with attention-based saliency scores, our method efficiently isolates the most significant semantic relationships for any given claim in a text.
  arXiv  Detail & Related papers  (2025-08-23T19:59:24Z)
• Self-Regularization with Sparse Autoencoders for Controllable LLM-based Classification [29.74457390987092]
  We propose a novel framework to identify and regularize unintended features in large language models (LLMs) latent spaces.<n>We evaluate the proposed framework on three real-world tasks, including toxic chat detection, reward modeling, and disease diagnosis.
  arXiv  Detail & Related papers  (2025-02-19T22:27:59Z)
• Appeal: Allow Mislabeled Samples the Chance to be Rectified in Partial Label Learning [55.4510979153023]
  In partial label learning (PLL), each instance is associated with a set of candidate labels among which only one is ground-truth. To help these mislabeled samples "appeal," we propose the first appeal-based framework.
  arXiv  Detail & Related papers  (2023-12-18T09:09:52Z)
• Robust Representation Learning for Unreliable Partial Label Learning [86.909511808373]
  Partial Label Learning (PLL) is a type of weakly supervised learning where each training instance is assigned a set of candidate labels, but only one label is the ground-truth. This is known as Unreliable Partial Label Learning (UPLL) that introduces an additional complexity due to the inherent unreliability and ambiguity of partial labels. We propose the Unreliability-Robust Representation Learning framework (URRL) that leverages unreliability-robust contrastive learning to help the model fortify against unreliable partial labels effectively.
  arXiv  Detail & Related papers  (2023-08-31T13:37:28Z)
• Ambiguity-Resistant Semi-Supervised Learning for Dense Object Detection [98.66771688028426]
  We propose a Ambiguity-Resistant Semi-supervised Learning (ARSL) for one-stage detectors. Joint-Confidence Estimation (JCE) is proposed to quantifies the classification and localization quality of pseudo labels. ARSL effectively mitigates the ambiguities and achieves state-of-the-art SSOD performance on MS COCO and PASCAL VOC.
  arXiv  Detail & Related papers  (2023-03-27T07:46:58Z)
• Human-Guided Fair Classification for Natural Language Processing [9.652938946631735]
  We show how to leverage unsupervised style transfer and GPT-3's zero-shot capabilities to generate semantically similar sentences that differ along sensitive attributes. We validate the generated pairs via an extensive crowdsourcing study, which confirms that a lot of these pairs align with human intuition about fairness in the context of toxicity classification.
  arXiv  Detail & Related papers  (2022-12-20T10:46:40Z)
• Textual Entailment Recognition with Semantic Features from Empirical Text Representation [60.31047947815282]
  A text entails a hypothesis if and only if the true value of the hypothesis follows the text. In this paper, we propose a novel approach to identifying the textual entailment relationship between text and hypothesis. We employ an element-wise Manhattan distance vector-based feature that can identify the semantic entailment relationship between the text-hypothesis pair.
  arXiv  Detail & Related papers  (2022-10-18T10:03:51Z)
• Measuring Fairness of Text Classifiers via Prediction Sensitivity [63.56554964580627]
  ACCUMULATED PREDICTION SENSITIVITY measures fairness in machine learning models based on the model's prediction sensitivity to perturbations in input features. We show that the metric can be theoretically linked with a specific notion of group fairness (statistical parity) and individual fairness.
  arXiv  Detail & Related papers  (2022-03-16T15:00:33Z)

This list is automatically generated from the titles and abstracts of the papers in this site.

This site does not guarantee the quality of this site (including all information) and is not responsible for any consequences.