OpenHAIV: A Framework Towards Practical Open-World Learning

2025-08-10

AI Summary

In open-world learning, existing out-of-distribution (OOD) detection methods lack mechanisms for knowledge updating, while incremental learning typically relies on supervised signals—introducing a critical mismatch with real-world open environments. To bridge this gap, we propose the first unified framework integrating OOD detection, novel-class discovery, and unsupervised incremental learning. Our approach leverages anomaly detection to identify unknown samples, employs contrastive learning coupled with clustering for unsupervised novel-class discovery, and introduces a gradient-constrained, prototype-aligned unsupervised fine-tuning strategy to ensure stable model evolution. Extensive experiments across multiple standard benchmarks demonstrate significant improvements in unknown-class recognition accuracy and long-term learning stability. Notably, our method achieves fully unsupervised open-world autonomous evolution—the first such result in the literature. Code and experimental artifacts are publicly released.

Technology Category

• Machine Learning: Unsupervised & Self-Supervised Learning
• Search and Optimization: Learning to Search
• Data Mining & Knowledge Management: Anomaly/Outlier Detection

Application Category

• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Graph Algorithms and Modeling for the Web: Foundation models and LLMs for Web-related graphs

Abstract

Substantial progress has been made in various techniques for open-world recognition. Out-of-distribution (OOD) detection methods can effectively distinguish between known and unknown classes in the data, while incremental learning enables continuous model knowledge updates. However, in open-world scenarios, these approaches still face limitations. Relying solely on OOD detection does not facilitate knowledge updates in the model, and incremental fine-tuning typically requires supervised conditions, which significantly deviate from open-world settings. To address these challenges, this paper proposes OpenHAIV, a novel framework that integrates OOD detection, new class discovery, and incremental continual fine-tuning into a unified pipeline. This framework allows models to autonomously acquire and update knowledge in open-world environments. The proposed framework is available at https://haiv-lab.github.io/openhaiv .

Problem

• Integrates OOD detection and incremental learning for open-world scenarios
• Enables autonomous knowledge acquisition and updates in open-world environments
• Addresses limitations of supervised conditions in incremental fine-tuning

Innovation

• Integrates OOD detection and new class discovery
• Combines incremental continual fine-tuning pipeline
• Autonomous knowledge acquisition in open-world