飞浆平台上完成基于MNIST手写数字识别数据集，设计合适的前馈神经网络进行实验，并取得95%以上的准确率

以下是在飞浆平台上完成基于MNIST手写数字识别数据集，设计合适的前馈神经网络进行实验，并取得95%以上的准确率的步骤：

1. 导入必要的库和数据集

import paddle
import paddle.fluid as fluid
from paddle.fluid.dygraph import Linear
from paddle.fluid.optimizer import SGDOptimizer
from paddle.fluid.dygraph import to_variable

train_dataset = paddle.dataset.mnist.train()
test_dataset = paddle.dataset.mnist.test()

2. 定义神经网络模型

class MNIST(fluid.dygraph.Layer):
    def __init__(self, name_scope):
        super(MNIST, self).__init__(name_scope)
        self.fc1 = Linear(input_dim=784, output_dim=200, act='sigmoid')
        self.fc2 = Linear(input_dim=200, output_dim=100, act='sigmoid')
        self.fc3 = Linear(input_dim=100, output_dim=10, act='softmax')

    def forward(self, inputs):
        x = self.fc1(inputs)
        x = self.fc2(x)
        x = self.fc3(x)
        return x

3. 定义训练函数和测试函数

def train(model):
    model.train()
    epoch_num = 10
    batch_size = 64
    optimizer = SGDOptimizer(learning_rate=0.1)
    for epoch in range(epoch_num):
        for batch_id, data in enumerate(train_reader()):
            x_data = np.array([x[0] for x in data], dtype='float32').reshape(-1, 784)
            y_data = np.array([x[1] for x in data], dtype='int64').reshape(-1, 1)
            img = to_variable(x_data)
            label = to_variable(y_data)
            predicts = model(img)
            loss = fluid.layers.cross_entropy(predicts, label)
            avg_loss = fluid.layers.mean(loss)
            if batch_id % 100 == 0:
                print("epoch: {}, batch_id: {}, loss is: {}".format(epoch, batch_id, avg_loss.numpy()))
            avg_loss.backward()
            optimizer.minimize(avg_loss)
            model.clear_gradients()

def test(model):
    model.eval()
    accuracies = []
    for batch_id, data in enumerate(test_reader()):
        x_data = np.array([x[0] for x in data], dtype='float32').reshape(-1, 784)
        y_data = np.array([x[1] for x in data], dtype='int64').reshape(-1, 1)
        img = to_variable(x_data)
        label = to_variable(y_data)
        predicts = model(img)
        predictions = np.argmax(predicts.numpy(), axis=1).reshape(-1, 1)
        accuracy = np.mean(predictions == y_data)
        accuracies.append(accuracy)
    avg_accuracy = np.mean(accuracies)
    print("accuracy is: {}".format(avg_accuracy))

4. 训练和测试模型

with fluid.dygraph.guard():
    model = MNIST("mnist")
    train_reader = paddle.batch(train_dataset, batch_size=64)
    test_reader = paddle.batch(test_dataset, batch_size=64)
    train(model)
    test(model)

运行以上代码，即可在飞浆平台上完成基于MNIST手写数字识别数据集，设计合适的前馈神经网络进行实验，并取得95%以上的准确率。