Anomaly Detection for Hybrid Butterfly Subspecies via Probability Filtering

2025-04-02

AI Summary

This work addresses the challenge of poor generalization of models trained on hybrid butterfly subspecies to novel mimetic subspecies. We propose a zero-shot cross-subspecies anomaly detection method that requires no retraining. Methodologically, we are the first to adapt the BioCLIP vision-language model—pretrained on biological features—to mimetic subspecies identification. To mitigate intra-class confusion, we introduce a probabilistic filtering mechanism; to enhance robustness against phenotypic variation in mimicry, we incorporate color jittering augmentation. Leveraging morphological mimicry as a biologically grounded prior, our approach enables reliable generalization from a known hybrid subspecies A to an unseen mimetic subspecies B. Evaluated on the NSF-HDR Challenge development set, our method ranked second. The source code is publicly available.

Technology Category

• Machine Learning: Bio-inspired Learning
• Computer Vision: Multi-modal Vision
• Cognitive Modeling & Cognitive Systems: Analogy

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Detecting butterfly hybrids requires knowledge of the parent subspecies, and the process can be tedious when encountering a new subspecies. This study focuses on a specific scenario where a model trained to recognize hybrid species A can generalize to species B when B biologically mimics A. Since species A and B share similar patterns, we leverage BioCLIP as our feature extractor to capture features based on their taxonomy. Consequently, the algorithm designed for species A can be transferred to B, as their hybrid and non-hybrid patterns exhibit similar relationships. To determine whether a butterfly is a hybrid, we adopt proposed probability filtering and color jittering to augment and simulate the mimicry. With these approaches, we achieve second place in the official development phase. Our code is publicly available at https://github.com/Justin900429/NSF-HDR-Challenge.

Problem

• Detect hybrid butterfly subspecies using probability filtering
• Generalize model from species A to mimic species B
• Leverage BioCLIP for taxonomic feature extraction

Innovation

• BioCLIP feature extractor for taxonomy
• Probability filtering for hybrid detection
• Color jittering to simulate mimicry