PromptAL: Sample-Aware Dynamic Soft Prompts for Few-Shot Active Learning
2025/07/22 by Xiang Hui, Jinqiao Shi, Xiang, Hui +9
Computer Science · Social Sciences · Engineering · #Machine Learning and Algorithms #Innovative Teaching Methods #Analog and Mixed-Signal Circuit Design
paper · pdf · doi:10.48550/arxiv.2507.16424
Abstract
Active learning (AL) aims to optimize model training and reduce annotation costs by selecting the most informative samples for labeling. Typically, AL methods rely on the empirical distribution of labeled data to define the decision boundary and perform uncertainty or diversity estimation, subsequently identifying potential high-quality samples. In few-shot scenarios, the empirical distribution often diverges significantly from the target distribution, causing the decision boundary to shift away from its optimal position. However, existing methods overlook the role of unlabeled samples in enhancing the empirical distribution to better align with the target distribution, resulting in a suboptimal decision boundary and the selection of samples that inadequately represent the target distribution. To address this, we propose a hybrid AL framework, termed PromptAL (Sample-Aware Dynamic Soft Prompts for Few-Shot Active Learning). This framework accounts for the contribution of each unlabeled data point in aligning the current empirical distribution with the target distribution, thereby optimizing the decision boundary. Specifically, PromptAL first leverages unlabeled data to construct sample-aware dynamic soft prompts that adjust the model's predictive distribution and decision boundary. Subsequently, based on the adjusted decision boundary, it integrates uncertainty estimation with both global and local diversity to select high-quality samples that more accurately represent the target distribution. Experimental results on six in-domain and three out-of-domain datasets show that PromptAL achieves superior performance over nine baselines. Our codebase is openly accessible.
Citations
- DeepSeek-R1 incentivizes reasoning in LLMs through reinforcement learning
- The Llama 3 Herd of Models
- Prompt Tuning Strikes Back: Customizing Foundation Models with Low-Rank Prompt Adaptation
- A Systematic Survey of Prompt Engineering in Large Language Models: Techniques and Applications
- Efficient Memory Management for Large Language Model Serving with PagedAttention
- A Multitask, Multilingual, Multimodal Evaluation of ChatGPT on Reasoning, Hallucination, and Interactivity
- MEAL: Stable and Active Learning for Few-Shot Prompting
- A Survey of Active Learning for Natural Language Processing
- A Comprehensive Survey of Few-shot Learning: Evolution, Applications, Challenges, and Opportunities
- IDPG: An Instance-Dependent Prompt Generation Method
- PromptBERT: Improving BERT Sentence Embeddings with Prompts
- PPT: Pre-trained Prompt Tuning for Few-shot Learning
- Active Learning by Acquiring Contrastive Examples
- Knowledgeable Prompt-tuning: Incorporating Knowledge into Prompt Verbalizer for Text Classification
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing
- Pre-train, Prompt, and Predict: A Systematic Survey of Prompting Methods in Natural Language Processing
- On Training Instance Selection for Few-Shot Neural Text Generation
- LoRA: Low-Rank Adaptation of Large Language Models
- SimCSE: Simple Contrastive Learning of Sentence Embeddings
- SimCSE: Simple Contrastive Learning of Sentence Embeddings
- GPT Understands, Too
- Calibrate Before Use: Improving Few-Shot Performance of Language Models
- Cold-start Active Learning through Self-supervised Language Modeling
- LEGAL-BERT: The Muppets straight out of Law School
- Exploiting Cloze Questions for Few Shot Text Classification and Natural\n Language Inference
- Learning the Difference that Makes a Difference with Counterfactually-Augmented Data
- Language Models as Knowledge Bases?
- RoBERTa: A Robustly Optimized BERT Pretraining Approach
- Diverse mini-batch Active Learning
- Weakly-Supervised Hierarchical Text Classification
- Character-level Convolutional Networks for Text Classification
- Bayesian Active Learning for Classification and Preference Learning
- A sequential algorithm for training text classifiers
- Multidimensional binary search trees used for associative searching
- BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding
- Divergence measures based on the Shannon entropy
- Nearest neighbor pattern classification
Related