ELM 分类器对二元分类给出零精度

Question

我的数据集（我们进行二进制分类的网络流量数据集）-

特征数为 25，我已经对数据集进行了标准化。

我的 ELM 模型-

class ELM:
    def __init__(self, num_input_nodes, num_hidden_units, num_out_units, activation='sigmoid',
                 loss='bce', beta_init=None, w_init=None, bias_init=None):
        self._num_input_nodes = num_input_nodes
        self._num_hidden_units = num_hidden_units
        self._num_out_units = num_out_units

        self._activation = getActivation(activation)
        self._loss = getLoss(loss)

        if isinstance(beta_init, np.ndarray):
            self._beta = beta_init
        else:
            self._beta = np.random.uniform(-1., 1., size=(self._num_hidden_units, self._num_out_units))

        if isinstance(w_init, np.ndarray):
            self._w = w_init
        else:
            self._w = np.random.uniform(-1, 1, size=(self._num_input_nodes, self._num_hidden_units))

        if isinstance(bias_init, np.ndarray):
            self._bias = bias_init
        else:
            self._bias = np.zeros(shape=(self._num_hidden_units,))

        print('Bias shape:', self._bias.shape)
        print('W shape:', self._w.shape)
        print('Beta shape:', self._beta.shape)

    def fit(self, X, Y, display_time=False):
        H = self._activation(X.dot(self._w) + self._bias)

        # Moore–Penrose pseudo inverse
        if display_time:
            start = time.time()
        H_pinv = np.linalg.pinv(H)
        if display_time:
            stop = time.time()
            print(f'Train time: {stop-start}')

        self._beta = H_pinv.dot(Y)

        # print('Fit Beta shape:', self._beta.shape)

    def __call__(self, X):
        H = self._activation(X.dot(self._w) + self._bias)
        return H.dot(self._beta)

    def evaluate(self, X, Y):
        pred = self(X)

        # Loss (base on model setting)
        loss = self._loss(Y, pred)

        # Accuracy
        acc = np.sum(np.argmax(pred, axis=-1) == np.argmax(Y, axis=-1)) / len(Y)

        # Unweighted Average Recall
        # TODO

        return loss, acc

# Network Settings
num_classes = 1
num_hidden_layers = 512
input_length = 25

当我尝试为我的数据集运行这个时，准确度变为零。我将 sigmoid 作为激活函数，将二元交叉熵作为我的二元分类任务的损失。