用集成对抗训练提升孟加拉语仇恨言论识别，低资源下表现优异。
Gradient Masters at BLP-2025 Task 1: Advancing Low-Resource NLP for Bengali using Ensemble-Based Adversarial Training for Hate Speech Detection
- 基于集成的对抗训练策略，增强模型对孟加拉语仇恨言论的判别能力。
- 在子任务1A和1B上分别取得73.23%和73.28%的微F1分数，排名第六与第三。
- 分析了误分类模式，为低资源语言仇恨内容检测提供可解释性支持。
本文针对BLP-2025任务1：'孟加拉多任务仇恨言论识别共享任务'，提出名为'Gradient Masters'的方法。针对YouTube评论中的子任务1A（仇恨类型分类）与1B（目标群体分类），采用基于集成的微调策略。在孟加拉语语言模型基础上设计混合方法，在低资源条件下显著优于基线模型。在子任务1A中以73.23%的微F1得分位列第6，在子任务1B中以73.28%得分位列第3。通过大量实验评估模型在开发与评估阶段的鲁棒性，对比不同语言模型变体，验证其在低资源孟加拉语仇恨言论场景下的泛化能力与数据集覆盖度。同时提供详尽的分析，探讨仇恨言论检测中的误分类模式。
原文摘要 · Abstract (English)
This paper introduces the approach of "Gradient Masters" for BLP-2025 Task 1: "Bangla Multitask Hate Speech Identification Shared Task". We present an ensemble-based fine-tuning strategy for addressing subtasks 1A (hate-type classification) and 1B (target group classification) in YouTube comments. We propose a hybrid approach on a Bangla Language Model, which outperformed the baseline models and secured the 6th position in subtask 1A with a micro F1 score of 73.23% and the third position in subtask 1B with 73.28%. We conducted extensive experiments that evaluated the robustness of the model throughout the development and evaluation phases, including comparisons with other Language Model variants, to measure generalization in low-resource Bangla hate speech scenarios and data set coverage. In addition, we provide a detailed analysis of our findings, exploring misclassification patterns in the detection of hate speech.
Thank you to arXiv for use of its open access interoperability. PaperDance 不是 arXiv 官方产品；中文卡片由大模型生成，请以原文为准。