Email Spam is a variety of automated spam where unbidden messages, used for business purpose, sent extensively to multiple mailing lists, individuals or newsgroups. To build a fruitful system for spam detection, we introduced Random Forest integrated with Deep Neural network to find the classification accuracy. The Random Forest algorithm uses a preordained probability of attributes in constructing their decision trees. The Gini measure is examined to rank the important features. The main objective is to grade the features using RF algorithm and to train the data using Deep Neural Network Classifier. Deep Neural Network Classifier model (DNNs) are trained using backpropagation algorithm in batch learning mode, which requires the entire training data to learn at once. The detector process was dynamically fit to the new data patterns till it reaches the spam coverage. Experimental results shows that classification rate of DNN is higher than compared to KNN and Support Vector Machine(SVM) with an accuracy of 88.59% while considering the top ranked five features.

电子邮件垃圾邮件是各种自动垃圾邮件，用于业务目的的未经禁止的邮件被广泛发送到多个邮件列表，个人或新闻组。为了构建一个有效的垃圾邮件检测系统，我们引入了与深度神经网络集成的随机森林以发现分类准确性。随机森林算法使用预先确定的属性概率来构造其决策树。对基尼度量进行了检查，以对重要特征进行排名。主要目的是使用RF算法对功能进行分级，并使用深度神经网络分类器训练数据。深度学习神经网络分类器模型（DNN）使用反向传播算法在批处理学习模式下进行训练，这需要立即学习整个训练数据。检测器进程可以动态适应新的数据模式，直到达到垃圾邮件覆盖率为止。实验结果表明，在考虑排名前五的特征时，DNN的分类率高于KNN和支持向量机（SVM），准确率达88.59％。