[1]
Z. Li, “Rethinking the Optimization of Average Precision: Only Penalizing Negative Instances before Positive Ones Is Enough”, AAAI, vol. 36, no. 2, pp. 1518-1526, Jun. 2022.