LabelRank算法(Python实现)

算法的定义和介绍就不收了，论文中都有，下面直接上代码实现。

def label_rank(A, in_, q, r):
    def ismember(a,b):
        for i in a:
            res = 1
            for j in b:
                if i not in j:
                    res = 0
                    break
            if res == 1:
                break
        return res
    
    Aori = A.copy()
    A = A + np.eye(len(A)) #添加自环
    # P_0 = A / A.sum(axis=0)
    # P_0 = np.array([[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0], [0, 0],[0, 0],[0, 0],[0, 1],[0, 0],[1, 0],[0, 0],[0, 0],
    # [0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0],[0, 0]])
    P_0 = np.eye(len(A))
    P_pre = np.dot(A.T, P_0)
    P_pre = P_pre / P_pre.sum(axis=0)
    a = 1
    COM = {}
    count = 0
    while a:
        #Step1:Propagation
        P_now = np.dot(A.T, P_pre)
        #Step2:Inflation
        P_in = np.power(P_now, in_)
        P_now = P_in / (P_in.sum(axis=1)+0.00001).reshape(-1, 1)
        #Step3:Cutoff
        P_now[P_now < r] = 0
        # #Step4:Explicit Conditional Update
        # MaNow = np.nanargmax(P_now, axis=1)
        # MaPre = np.nanargmax(P_pre, axis=1)
        # restart = []
        # for i in range(len(A)):
        #     gain = 0
        #     Nb = np.nonzero(Aori[i,:])
        #     MaxI = np.nanargmax(P_now[i, :])
        #     MaxI = np.nonzero(P_now[i, :] == MaxI)
        #     MaxNb = MaNow[Nb]
        #     for k in range(len(Nb)):
        #         MaxNbID = np.nonzero(P_now[Nb[k], :] == MaxNb[k])
        #         if ismember(MaxI, MaxNbID):
        #             gain += 1
        #     if gain >= q * len(Nb):
        #         restart.append(i)
        # P_now[restart, :] = P_pre[restart, :]
        #Step5:Stop Criterion
        # if ismember(np.nonzero(P_now[i, :]==MaNow[i]), np.nonzero(P_pre[i, :] == MaPre[i])):
        if count == 50:
            a = 0
        P_pre = P_now
        count += 1
    R = P_now
    # for one in range(len(P_now)):
    #     print(one+1, P_now[one,:])
    return R

以上是LabelRank算法的Python实现版本，目前来说还比较粗糙，还有许多地方需要优化完善，后续版本敬请期待。