CLR, a novel contrastive learning method using graph-based sample relationships. This approach outperformed traditional ...

Available methods to randomize binary matrices with fixed row and column sums are computationally intensive and tend to generate matrix configurations with unequal frequency.