检测用户命令序列异常——使用LSTM分类算法【使用朴素贝叶斯，类似垃圾邮件分类的做法也可以，将命令序列看成是垃圾邮件】

通过 搜集 Linux 服务器 的 bash 操作 日志， 通过 训练 识别 出 特定 用户 的 操作 习惯， 然后 进一步 识别 出 异常 操作 行为。

使用 SEA 数据 集 涵盖 70 多个 UNIX 系统 用户 的 行为 日志， 这些 数据 来自 UNIX 系统 acct 机制 记录 的 用户 使用 的 命令。 SEA 数据 集中 每个 用户 都 采集 了 15000 条 命令， 从 用户 集合 中 随机 抽取 50 个 用户 作为 正常 用户， 剩余 用户 的 命令 块 中 随机 插入 模拟 命令 作为 内部 伪装 者 攻击 数据。其中 训练 集合 大小 为 80， 测试 集合 大小 为 70。

数据集示意：

cpp
sh
xrdb
cpp
sh
xrdb
mkpts
test
stty
hostname
date
echo
[
find
chmod
tty
echo
env
echo
sh
userenv
wait4wm
xhost
xsetroot
reaper
xmodmap
sh
[
cat
stty
hostname
date
echo
[
find
chmod
tty
echo
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
more
sh
launchef
launchef
sh
9term
sh
launchef
sh
launchef
hostname
[
cat
stty
hostname
date
echo
[
find
chmod
tty
echo
sh
more
sh
more
sh
ex
sendmail
sendmail
sh
MediaMai
sendmail
sh
rm
MediaMai
sh
rm
MediaMai
launchef
launchef
sh
sh
more
sh
sh
rm
MediaMai
netstat
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
netscape
sh
netscape
more
sh
rm
sh
MediaMai
=
telnet
tput
netscape
netscape
netscape
netscape
netscape

# -*- coding:utf-8 -*-

import sys

import re
import numpy as np

import nltk
import csv
import matplotlib.pyplot as plt
from nltk.probability import FreqDist
from sklearn.feature_extraction.text import CountVectorizer

from sklearn import cross_validation
from tflearn.data_utils import to_categorical, pad_sequences
from tflearn.datasets import imdb
import tflearn

#测试样本数
N=80

def load_user_cmd_new(filename):
    cmd_list=[]
    dist=[]
    with open(filename) as f:
        i=0
        x=[]
        for line in f:
            line=line.strip('\n')
            x.append(line)
            dist.append(line)
            i+=1
            if i == 100:
                cmd_list.append(x)
                x=[]
                i=0

    fdist = FreqDist(dist).keys()
    return cmd_list,fdist

def load_user_cmd(filename):
    cmd_list=[]
    dist_max=[]
    dist_min=[]
    dist=[]
    with open(filename) as f:
        i=0
        x=[]
        for line in f:
            line=line.strip('\n')
            x.append(line)
            dist.append(line)
            i+=1
            if i == 100:
                cmd_list.append(x)
                x=[]
                i=0

    fdist = FreqDist(dist).keys()
    dist_max=set(fdist[0:50])
    dist_min = set(fdist[-50:])
    return cmd_list,dist_max,dist_min

def get_user_cmd_feature(user_cmd_list,dist_max,dist_min):
    user_cmd_feature=[]
    for cmd_block in user_cmd_list:
        f1=len(set(cmd_block))
        fdist = FreqDist(cmd_block).keys()
        f2=fdist[0:10]
        f3=fdist[-10:]
        f2 = len(set(f2) & set(dist_max))
        f3=len(set(f3)&set(dist_min))
        x=[f1,f2,f3]
        user_cmd_feature.append(x)
    return user_cmd_feature

def get_user_cmd_feature_new(user_cmd_list,dist):
    user_cmd_feature=[]
    for cmd_list in user_cmd_list:
        x=[]
        for cmd in  cmd_list:
            v = [0] * len(dist)
            for i in range(0, len(dist)):
                if cmd == dist[i]:
                    v[i] = 1
            x.append(v)
        user_cmd_feature.append(x)
    return user_cmd_feature

def get_label(filename,index=0):
    x=[]
    with open(filename) as f:
        for line in f:
            line=line.strip('\n')
            x.append( int(line.split()[index]))
    return x

def do_knn(x_train,y_train,x_test,y_test):
    neigh = KNeighborsClassifier(n_neighbors=3)
    neigh.fit(x_train, y_train)
    y_predict=neigh.predict(x_test)
    score = np.mean(y_test == y_predict) * 100

    print  score

def do_rnn(x_train,x_test,y_train,y_test):
    global n_words
    # Data preprocessing
    # Sequence padding
    print "GET n_words embedding %d" % n_words

    #x_train = pad_sequences(x_train, maxlen=100, value=0.)
    #x_test = pad_sequences(x_test, maxlen=100, value=0.)
    # Converting labels to binary vectors
    y_train = to_categorical(y_train, nb_classes=2)
    y_test = to_categorical(y_test, nb_classes=2)

    # Network building
    net = tflearn.input_data(shape=[None, 100,n_words])
    net = tflearn.lstm(net, 10,  return_seq=True)
    net = tflearn.lstm(net, 10, )
    net = tflearn.fully_connected(net, 2, activation='softmax')
    net = tflearn.regression(net, optimizer='adam', learning_rate=0.1,name="output",
                             loss='categorical_crossentropy')

    # Training

    model = tflearn.DNN(net, tensorboard_verbose=3)
    model.fit(x_train, y_train, validation_set=(x_test, y_test), show_metric=True,
             batch_size=32,run_id="maidou")

if __name__ == '__main__':
    user_cmd_list,dist=load_user_cmd_new("../data/MasqueradeDat/User7")
    #print  "Dist:(%s)" % dist
    n_words=len(dist)
    user_cmd_feature=get_user_cmd_feature_new(user_cmd_list,dist)

    labels=get_label("../data/MasqueradeDat/label.txt",6)
    y=[0]*50+labels

    x_train=user_cmd_feature[0:N]
    y_train=y[0:N]

    x_test=user_cmd_feature[N:150]
    y_test=y[N:150]

    #print x_train

    do_rnn(x_train,x_test,y_train,y_test)

效果：

Training Step: 30  | total loss: 0.10088 | time: 1.185s
| Adam | epoch: 010 | loss: 0.10088 - acc: 0.9591 | val_loss: 0.18730 - val_acc: 0.9571 -- iter: 80/80
--