Bio-invasion: A prediction model based on multi-objective optimization

In some real application scenarios, it is difficult to obtain enough labeled data and get an ideal classification result. Unsupervised learning and few-shot classification are widely used for this kind of tasks. In contrast to prior approaches, we propose a multi-objective optimization method, combining both location and image information. We establish models for these two kinds of information and calculate Pareto frontier in multi-objective programming, which enhances the interpretability with clear evaluation indicators. Our method is applied in the detection on Asian giant hornet and attains favorable results across few-shot datasets.