AutoAL: Automated Active Learning with Differentiable Query Strategy Search

2024-10-17

arXiv.org

AI Summary

Selecting optimal active learning (AL) strategies is challenging under high annotation costs due to the combinatorial and non-differentiable nature of conventional AL heuristics. Method: We propose the first differentiable AL strategy search framework, formulating strategy discovery as a differentiable bi-level optimization problem. It jointly trains SearchNet—generating parameterized AL strategies—and FitNet—estimating their downstream performance—enabling end-to-end, task-adaptive strategy discovery. Inspired by neural architecture search and meta-learning, our approach constructs a differentiable compositional strategy space grounded in existing base samplers. Contribution/Results: Evaluated across multiple benchmark datasets, our framework consistently outperforms all baseline AL methods under identical annotation budgets. It demonstrates strong cross-task generalization and plug-and-play applicability without task-specific re-design, establishing a new paradigm for automated, differentiable AL strategy optimization.

Technology Category

• Search and Optimization: Learning to Search
• Machine Learning: Active Learning
• Constraint Satisfaction and Optimization: Search

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Economics, Online Markets and Human Computation: Humans versus LLMs for data annotation and labeling
• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs

Abstract

As deep learning continues to evolve, the need for data efficiency becomes increasingly important. Considering labeling large datasets is both time-consuming and expensive, active learning (AL) provides a promising solution to this challenge by iteratively selecting the most informative subsets of examples to train deep neural networks, thereby reducing the labeling cost. However, the effectiveness of different AL algorithms can vary significantly across data scenarios, and determining which AL algorithm best fits a given task remains a challenging problem. This work presents the first differentiable AL strategy search method, named AutoAL, which is designed on top of existing AL sampling strategies. AutoAL consists of two neural nets, named SearchNet and FitNet, which are optimized concurrently under a differentiable bi-level optimization framework. For any given task, SearchNet and FitNet are iteratively co-optimized using the labeled data, learning how well a set of candidate AL algorithms perform on that task. With the optimal AL strategies identified, SearchNet selects a small subset from the unlabeled pool for querying their annotations, enabling efficient training of the task model. Experimental results demonstrate that AutoAL consistently achieves superior accuracy compared to all candidate AL algorithms and other selective AL approaches, showcasing its potential for adapting and integrating multiple existing AL methods across diverse tasks and domains. Code will be available at: https://github.com/haizailache999/AutoAL.

Problem

• Automating optimal active learning strategy selection for diverse tasks
• Reducing labeling costs via differentiable query strategy search
• Improving deep learning efficiency with adaptive AL integration

Innovation

• Differentiable query strategy search method
• Bi-level optimization with SearchNet and FitNet
• Automated selection of optimal AL strategies