近年来恶意软件猖獗，解决其问题刻不容缓。虽用机器学习进行恶意软件分类已较成熟，但它对特征处理要求高，因此我采用深度学习思路，基于Canadian Institute for Cybersecurity提供的近年安卓恶意软件提取特征文件进行改进，初步完成了一个入门级恶意软件分类项目。

[网络安全篇4] 全连接网络分类实现

近年来，恶意软件的出现尤为猖獗。所以解决恶意软件的问题也成为了刻不容缓的事情，当然，现在运用机器学习去完成恶意软件分类的手段已经比较的成熟了，但是，机器学习去特征的处理要求是比较高的，这里我采用深度学习的思维去解决。与此同时，Canadian Institute for Cybersecurity这个研究所提供了近几年的安卓恶意软件的提取特征文件，我在此基础上进行改进，初步完成一个较为入门级别的恶意软件分类项目。

0 背景

前面使用了图像的思想去进行分类，发现效果并不好，单纯的采用机器学习的方法都可以达到一个比较不错的结果了，那我们现在就摒弃图像的想法，继续探索正常的分类，其实，一排串的数据是一类，我们其实就可以把他当成文本分类的概念去解决。这样的话，就慢慢的将解决思路转向NLP，然后就可以直接去使用transformer的那些套路完成项目，当然奔着循序渐进的理念，本项目不会涉及transformer，先用全连接进行一个网络搭建。

1 数据集

详情参照CIC研究所的具体发布内容。

https://www.unb.ca/cic/datasets/maldroid-2020.html

这里我采用了其中470维度的一个数据进行处理，数据集已经挂载在项目数据集当中。

# 对数据进行一个copy!cp data/data157200/syscallsbinders_frequency_5_Cat.csv work/登录后复制 ? ?

2 数据处理

2.1 数据展示

因为是csv文件，我这里采用pandas的方式进行读取，更为方便快速。

可以通过表头看到，是各种权限、调用函数、绑定器的使用频次统计。

每一行的最后一列是该行数据所属于的分类类别。

import pandas as pd# 按文件名读取整个文件data_471 = pd.read_csv(”work/syscallsbinders_frequency_5_Cat.csv“) data_471.head(2) # 展示前几行数据，通常为展示前5行登录后复制 ? ? ? ?

ACCESS_PERSONAL_INFO___ ALTER_PHONE_STATE___ ANTI_DEBUG_____ 1 0 0 1 3 0 0 CREATE_FOLDER_____ CREATE_PROCESS`_____ CREATE_THREAD_____ 3 0 14 1 6 0 42 DEVICE_ACCESS_____ EXECUTE_____ FS_ACCESS____ FS_ACCESS()____ ... 2 0 3 0 ... 1 91 0 32 0 ... utimes vfork vibrate vibratePattern wait4 watchRotation 0 0 0 0 0 0 1 0 0 0 0 0 0 windowGainedFocus write writev Class 0 0 37 10 1 1 2 2838 46 1 [2 rows x 471 columns]登录后复制 ? ? ? ? ? ? ? ?

2.2 无效数据过滤

我这里的思维就是，竖向的每一列，如果求和他的值为0，那么他的这一列的数值存在将是毫无意义的。

换句话来讲，其实这里也是一种特征降维。

前面的图像解决有可以的去将数据控制到一个$NxN+1$NxN+1的维度,这里就不再需要这样处理了

我这里只把为全为0的列去除。

（发现并没有全为0的列QAQ）

import pandas as pd# 按文件名读取整个文件data_471 = pd.read_csv(”work/syscallsbinders_frequency_5_Cat.csv“) df = pd.DataFrame(data_471)len(df.columns)# 记录每一列求和小于阈值的列threshold = 0de = []# 记录样本数据中的最大值de_max = 0de_va = []for i in range(0,470): name = df.columns.values[i] if(df[name].sum() < threshold): de.append(name) if(df[name].max() >= de_max): de_max = df[name].max()# 删除小于阈值的列for j in de: df.drop(j,axis=1,inplace=True)len_new = len(df.columns)len_new登录后复制 ? ? ? ?

471登录后复制 ? ? ? ? ? ? ? ?In [?]

de_max登录后复制 ? ? ? ?

3697410登录后复制 ? ? ? ? ? ? ? ?In [?]

# 新数据写入csv 虽然没有变化，新存一下吧，养成一个良好的文件备份习惯df.to_csv(”new_date_“+str(len_new)+”.csv“,index=False,sep=',')登录后复制 ? ?In [?]

import numpy as npimport pandas as pdmydata = pd.read_csv(”new_date_“+str(len_new)+”.csv“)row_list = mydata.values.tolist()# 一个数据展示print(row_list[0])登录后复制 ? ? ? ?

[1, 0, 0, 3, 0, 14, 2, 0, 3, 0, 3, 0, 0, 0, 4, 0, 9, 0, 7, 2, 0, 0, 0, 0, 0, 0, 0, 0, 3, 0, 4, 0, 0, 0, 3, 0, 14, 6, 0, 0, 11, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 3, 21, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 4, 0, 0, 109, 14, 24, 0, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 3, 0, 0, 3, 0, 0, 0, 0, 0, 20, 0, 0, 0, 0, 2, 0, 14, 0, 2, 0, 0, 170, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 7, 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, 2, 0, 1, 0, 0, 0, 0, 0, 0, 0, 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, 2, 0, 0, 0, 0, 0, 0, 18, 0, 23, 0, 0, 0, 6, 6, 33, 26, 25, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 70, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 5, 61, 3, 0, 0, 61, 0, 133, 0, 0, 0, 22, 8, 0, 0, 0, 0, 0, 0, 0, 0, 0, 28, 0, 0, 0, 0, 2, 0, 0, 0, 0, 4, 96, 12, 0, 0, 0, 0, 0, 0, 0, 0, 29, 0, 0, 3, 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, 2, 0, 0, 0, 0, 0, 0, 4, 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, 1, 0, 0, 14, 1, 1, 1, 0, 9, 0, 0, 0, 0, 1, 17, 31, 0, 7, 0, 0, 0, 0, 0, 0, 20, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 37, 10, 1]登录后复制 ? ? ? ?In [?]

len(row_list)登录后复制 ? ? ? ?

11598登录后复制 ? ? ? ? ? ? ? ?In [?]

# 获取最后一维的数据，也就是分类类别 1-5 这个是一行的分类的展示row_list[0][-1]登录后复制 ? ? ? ?

1登录后复制 ? ? ? ? ? ? ? ?In [?]

import numpy as np# 获取类别分布 所有行的最后一列为labelcls_ = np.asarray(mydata.iloc[:-2,-1])print('Labels:', np.unique(cls_))登录后复制 ? ? ? ?

Labels: [1 2 3 4 5]登录后复制 ? ? ? ?In [?]

_records = { i: sum([1 for j in cls_ if j==i]) for i in range(1,6)}print('dist:', _records)登录后复制 ? ? ? ?

dist: {1: 1253, 2: 2100, 3: 3904, 4: 2546, 5: 1793}登录后复制 ? ? ? ?

2.3 控制划分后数据类别

防止划分后打乱的数据分类不均衡

import randomrandom.shuffle(row_list)登录后复制 ? ?In [?]

# 验证一下是不是真的打乱了 因为最原始的数据前面全是1分类，所以下面的输出只要不是全1 就是打乱的了for i in range(4): print(row_list[i][-1])登录后复制 ? ? ? ?

3435登录后复制 ? ? ? ?In [?]

# 划分数据train_rate = 0.8train_list = []eval_list = []_cls_records = {i:0 for i in cls_}for data in row_list: _data_str, _label_int= data[:-1],data[-1:] label = int(_label_int[0]) # 控制每个类别的比例 if _cls_records[label] < int(_records[label]*train_rate): train_list.append([_data_str,_label_int]) else: eval_list.append([_data_str,_label_int]) _cls_records[label] += 1print(len(train_list))print(len(eval_list))print(len(train_list)+len(eval_list))登录后复制 ? ? ? ?

9275232311598登录后复制 ? ? ? ?In [?]

# 解析数据样本train_data = []train_label = []for data_label in train_list: data,label = data_label[0],data_label[1] train_data.append(data) train_label.append(label)val_data = []val_label = []for data_label in eval_list: data,label = data_label[0],data_label[1] val_data.append(data) val_label.append(label)登录后复制 ? ?In [?]

print(”Train Dataset Length:“, len(train_data))print(”Eval Dataset Length:“, len(val_data))print(”Example Sample:“, train_data[0], '------', train_label[0])登录后复制 ? ? ? ?

Train Dataset Length: 9275Eval Dataset Length: 2323Example Sample: [2, 0, 0, 3, 0, 25, 4, 0, 9, 0, 2, 0, 0, 2, 2, 0, 14, 0, 7, 4, 0, 0, 0, 4, 0, 8, 0, 0, 1, 0, 1, 0, 0, 0, 1, 0, 25, 5254, 0, 0, 18, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 1, 70, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 2, 0, 0, 446, 25, 27, 0, 1, 0, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 7, 11, 59, 0, 0, 1, 0, 0, 0, 0, 0, 69, 5, 0, 0, 0, 2, 0, 31, 0, 0, 0, 0, 440, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 7, 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, 3, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 2, 10, 10, 10, 0, 0, 0, 36, 0, 30, 0, 0, 0, 0, 3, 177, 10, 40, 0, 0, 0, 0, 1, 0, 3, 0, 0, 0, 0, 0, 0, 91, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 1, 0, 36, 10, 1034, 3, 0, 0, 1029, 0, 285, 0, 0, 0, 14, 7, 0, 0, 0, 0, 0, 0, 0, 0, 0, 40, 0, 0, 0, 0, 8, 0, 0, 0, 0, 0, 1048, 9, 0, 0, 0, 0, 0, 0, 0, 0, 5311, 0, 0, 0, 2, 0, 1, 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, 2, 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, 1, 0, 0, 26, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 26, 21, 0, 2, 8, 0, 0, 0, 0, 0, 19, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 2, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 102, 30] ------ [3]登录后复制 ? ? ? ?

到这里，数据就处理完了，就可以编写数据处理类和模型网络搭建了。

3 dataset类编写

from paddle.io import Datasetfrom PIL import Imageclass MyDataset(Dataset): def __init__(self, mode = 'train'): # 训练样本数量 self.training_data, self.training_label, self.test_data, self.test_label = train_data, train_label, val_data, val_label if mode == 'train': self.num_samples = len(train_data) else: self.num_samples = len(val_data) self.mode = mode def __getitem__(self, idx): if self.mode == 'train': data = self.training_data[idx] label = self.training_label[idx] else: data = self.test_data[idx] label = self.test_label[idx] data = np.array(data).astype('float32') return data, np.array(label, dtype='int64') def __len__(self): # 返回样本总数量 return self.num_samples登录后复制 ? ?In [?]

# 训练的数据提供器train_dataset = MyDataset(mode='train')# 测试的数据提供器eval_dataset = MyDataset(mode='val')# 查看训练和测试数据的大小print('train大小：', train_dataset.__len__())print('eval大小：', eval_dataset.__len__())# 查看图片数据、大小及标签for data, label in train_dataset: print(np.array(data).shape) print(np.array(label).shape) break登录后复制 ? ? ? ?

train大小： 9275eval大小： 2323(470,)(1,)登录后复制 ? ? ? ?

4 全连接网络搭建

4.1 模型搭建

import paddleimport paddle.nn as nn# 实现MLP模块class Mlp(nn.Layer): # self代表类的实例自身 def __init__(self): # 初始化父类中的一些参数 super(Mlp, self).__init__() self.fc1 = paddle.nn.Linear(in_features=470, out_features=1024) self.bn1 = paddle.nn.BatchNorm1D(1024) self.relu1 = paddle.nn.ReLU() self.fc2 = paddle.nn.Linear(in_features=1024, out_features=512) self.bn2 = paddle.nn.BatchNorm1D(512) self.relu2 = paddle.nn.ReLU() self.fc3 = paddle.nn.Linear(in_features=512, out_features=64) self.bn3 = paddle.nn.BatchNorm1D(64) self.relu3 = paddle.nn.ReLU() self.fc4 = paddle.nn.Linear(in_features=64, out_features=1) # 网络的前向计算 def forward(self, inputs): x = self.fc1(inputs) x = self.bn1(x) x = self.relu1(x) x = self.fc2(x) x = self.bn2(x) x = self.relu2(x) x = self.fc3(x) x = self.bn3(x) x = self.relu3(x) x = self.fc4(x) return x登录后复制 ? ?In [?]

# 测试mlp = Mlp()# 模型结构展示paddle.summary(mlp, (256, 470))登录后复制 ? ? ? ?

--------------------------------------------------------------------------- Layer (type) Input Shape Output Shape Param # =========================================================================== Linear-77 [[256, 470]] [256, 1024] 482,304 BatchNorm1D-58 [[256, 1024]] [256, 1024] 4,096 ReLU-58 [[256, 1024]] [256, 1024] 0 Linear-78 [[256, 1024]] [256, 512] 524,800 BatchNorm1D-59 [[256, 512]] [256, 512] 2,048 ReLU-59 [[256, 512]] [256, 512] 0 Linear-79 [[256, 512]] [256, 64] 32,832 BatchNorm1D-60 [[256, 64]] [256, 64] 256 ReLU-60 [[256, 64]] [256, 64] 0 Linear-80 [[256, 64]] [256, 1] 65 ===========================================================================Total params: 1,046,401Trainable params: 1,040,001Non-trainable params: 6,400---------------------------------------------------------------------------Input size (MB): 0.46Forward/backward pass size (MB): 9.38Params size (MB): 3.99Estimated Total Size (MB): 13.83---------------------------------------------------------------------------登录后复制 ? ? ? ?

{'total_params': 1046401, 'trainable_params': 1040001}登录后复制 ? ? ? ? ? ? ? ?

4.2 模型训练

import paddlefrom paddle.io import DataLoaderBATCH_SIZE = 256 # 设置batch大小# 训练的数据提供器train_dataloader = DataLoader( dataset = train_dataset, batch_size = BATCH_SIZE, shuffle = True, drop_last=True )# 测试的数据提供器eval_dataloader = DataLoader( dataset = eval_dataset, batch_size = BATCH_SIZE, shuffle =False, drop_last=True )登录后复制 ? ?In [73]

import osEPOCH_NUM = 1500 # 设置外层循环次数save_dir = 'output_model/'# 模型封装model = Mlp()learning_rate = 5e-3opt = paddle.optimizer.SGD(learning_rate=learning_rate, parameters=model.parameters())def train_mlp(): # 定义外层循环 train_loss = [] val_loss = [] for epoch_id in range(EPOCH_NUM): # 开启模型训练模式 model.train() for iter_id, mini_batch in enumerate(train_dataloader()): # 清空梯度变量，以备下一轮计算 opt.clear_grad() x = np.array(mini_batch[:][0]) y = np.array(mini_batch[:][1]) # 将numpy数据转为飞桨动态图tensor的格式 features = paddle.to_tensor(x,dtype='float32') y = paddle.to_tensor(y,dtype='float32') # 前向计算 predicts = model(features) # 计算损失 loss = paddle.nn.functional.l1_loss(predicts, label=y) avg_loss = paddle.mean(loss) train_loss.append(avg_loss.numpy()) # 反向传播，计算每层参数的梯度值 avg_loss.backward() # 更新参数，根据设置好的学习率迭代一步 opt.step() # 开启验证模式 model.eval() for iter_id, mini_batch in enumerate(eval_dataloader()): x = np.array(mini_batch[:][0]) y = np.array(mini_batch[:][1]) features = paddle.to_tensor(x,dtype='float32') y = paddle.to_tensor(y,dtype='float32') predicts = model(features) loss = paddle.nn.functional.l1_loss(predicts, label=y) avg_loss = paddle.mean(loss) val_loss.append(avg_loss.numpy()) print(f'Epoch {epoch_id}, train loss is {np.mean(train_loss)}, val loss is {np.mean(val_loss)}') if (epoch_id % 5 == 0): if not os.path.isdir(save_dir): os.makedirs(save_dir) save_path = os.path.join(save_dir, '{0}'.format(epoch_id)) paddle.save(model.state_dict(), save_path + '.pdparams') paddle.save(opt.state_dict(), save_path + '.pdopt') return train_loss,val_loss登录后复制 ? ?In [74]

tloss,vloss = train_mlp()登录后复制 ? ?

4.3 训练日志可视化

import matplotlib.pyplot as plt%matplotlib inlinex = [i for i in range(len(tloss))]y = [i for i in range(len(vloss))]def draw(x,tloss,label): plt.title(label,fontsize = 24) plt.xlabel('iter_id',fontsize = 20) plt.ylabel('loss',fontsize = 20) plt.plot(x,tloss,color = 'green',label = label) plt.legend() plt.grid() plt.show()draw(x,tloss,'train loss')draw(y,vloss,'val loss')登录后复制 ? ? ? ?

/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/matplotlib/__init__.py:107: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working from collections import MutableMapping/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/matplotlib/rcsetup.py:20: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working from collections import Iterable, Mapping/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/matplotlib/colors.py:53: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working from collections import Sized/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/matplotlib/cbook/__init__.py:2349: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working if isinstance(obj, collections.Iterator):/opt/conda/envs/python35-paddle120-env/lib/python3.7/site-packages/matplotlib/cbook/__init__.py:2366: DeprecationWarning: Using or importing the ABCs from 'collections' instead of from 'collections.abc' is deprecated, and in 3.8 it will stop working return list(data) if isinstance(data, collections.MappingView) else data登录后复制 ? ? ? ?

<Figure size 432x288 with 1 Axes>登录后复制登录后复制 ? ? ? ? ? ? ? ?

<Figure size 432x288 with 1 Axes>登录后复制登录后复制 ? ? ? ? ? ? ? ?

5 模型保存及简单验证

5.1 模型保存

paddle.save(model.state_dict(), 'out/final.pdparams') # 保存模型登录后复制 ? ?

5.2 模型验证

其实上面的可视化图已经是对结果的一个很好的验证了，这里我们通过实际的数据的展示，来看一下效果。

print(val_data[10])print(val_label[10][0])登录后复制 ? ? ? ?

[0, 0, 0, 3, 0, 9, 2, 0, 3, 0, 3, 0, 0, 0, 3, 0, 8, 0, 6, 1, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 1, 0, 9, 6, 0, 0, 6, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 20, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 3, 0, 0, 109, 9, 19, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 1, 31, 0, 0, 1, 0, 0, 0, 0, 0, 5, 0, 0, 0, 0, 2, 0, 8, 0, 1, 0, 0, 1184, 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, 5, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 12, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 1, 0, 0, 0, 0, 0, 0, 0, 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, 2, 0, 0, 0, 0, 0, 0, 27, 0, 30, 0, 0, 0, 0, 3, 23, 9, 46, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 103, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 8, 4, 47, 3, 0, 0, 35, 0, 183, 0, 0, 0, 5, 11, 0, 0, 0, 0, 0, 0, 0, 0, 0, 25, 0, 0, 0, 0, 2, 0, 0, 0, 0, 0, 91, 1, 0, 0, 0, 0, 0, 0, 0, 0, 24, 0, 0, 0, 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, 1, 0, 0, 0, 0, 0, 0, 1, 0, 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, 1, 0, 0, 6, 1, 1, 1, 0, 0, 0, 0, 0, 0, 1, 10, 1, 0, 1, 0, 0, 0, 0, 0, 0, 29, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 1, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 65, 3]2登录后复制 ? ? ? ?In [80]

import paddle.nn.functional as Flabel_txt = ['','Adware','Banking','SMS malware','Riskware','Benign']model = Mlp()para_dict = paddle.load('out/final.pdparams')model.load_dict(para_dict)model.eval()x = paddle.reshape(paddle.to_tensor(val_data[10],dtype='float32'),[-1,470])predicts = model(x)pre = paddle.round(predicts)print(”实际标签为:“,label_txt[val_label[10][0]])print(”预测标签为:“,label_txt[pre[0]])登录后复制 ? ? ? ?

实际标签为: Banking预测标签为: Banking登录后复制 ? ? ? ?