TABLE 1. Results from a in The clownfish-hosting sea anemones (Anthozoa: Actiniaria): updated nomenclature, biogeography, and practical field guide.

TABLE 1. Results from a machine learning neural network analysis to automate species identification for the clownfishhosting sea anemones. Sample size reflects the total number of digital photographs aggregated for this analysis. Precision represents the percent accuracy of the neural network in identifying a given species correctly.
SpeciesSample SizePrecisionRecallFl-scoreSupportCryptodendrum adhaesivum900.360.130.2030Entacmaea quadricolor9920.550.600.57310Heteractis aurora1500.210.190.2042Radianthus crispa2900.260.240.2574Radianthus doreensis1290.270.230.2544Radianthus magnifica10530.550.650.60302Radianthus malu340.000.000.0010Stichodactyla gigantea1350.380.290.3242Stichodactyla haddoni2040.470.370.4168Stichodactyla mertensii3170.280.200.2382Accuracy0.481004Macro avg.0.330.290.301004Weighted avg.0.460.480.461004

表1 用于自动化识别与小丑鱼共生的海葵物种的机器学习神经网络分析结果。样本量指本次分析所聚合的数码照片总数量。精确率（Precision）代表神经网络对某一物种的正确识别准确率百分比。

| 物种 | 样本量 | 精确率 | 召回率 | F1分数 | 支持样本数 |
|--------------------------|--------|--------|--------|--------|------------|
| Cryptodendrum adhaesivum | 90 | 0.36 | 0.13 | 0.20 | 30 |
| Entacmaea quadricolor | 992 | 0.55 | 0.60 | 0.57 | 310 |
| Heteractis aurora | 150 | 0.21 | 0.19 | 0.20 | 42 |
| Radianthus crispa | 290 | 0.26 | 0.24 | 0.25 | 74 |
| Radianthus doreensis | 129 | 0.27 | 0.23 | 0.25 | 44 |
| Radianthus magnifica | 1053 | 0.55 | 0.65 | 0.60 | 302 |
| Radianthus malu | 34 | 0.00 | 0.00 | 0.00 | 10 |
| Stichodactyla gigantea | 135 | 0.38 | 0.29 | 0.32 | 42 |
| Stichodactyla haddoni | 204 | 0.47 | 0.37 | 0.41 | 68 |
| Stichodactyla mertensii | 317 | 0.28 | 0.20 | 0.23 | 82 |
| 准确率 | — | 0.48 | — | — | 1004 |
| 宏平均（Macro avg.） | — | 0.33 | 0.29 | 0.30 | 1004 |
| 加权平均（Weighted avg.）| — | 0.46 | 0.48 | 0.46 | 1004 |