Applying Deep learning to Improve detection rate in IDS using Tensorflow

Machine learning focus tasks associated with classification and recognition, which are related to artificial intelligence science. Generally, these kinds of operations are performed using mathematical and heuristic approaches in other words, effective result can be captured from huge space solutions.

There are well known approaches in machine learning such as neural network, support vector machine, fuzzy logic, swarm intelligence and more. Despite the number of these approaches and algorithms, there are applications for which higher degree of accuracy and lower error rate are required. However, ANN or artificial neural network can be applied in malware detection or classification, face recognition, finger prints analysis in which data-set is used for training the model and then predication of further data is done. However, ANN is fully dependent on the data-set used for training the model, because of that if the data is not accurate and not normalized, the predictive analysis will affect the accuracy rate.

Tensorflow python library was used to develop and train the model. Tensorflow (Google Brain) is an open source software library used for deep learning. It is based on set of algorithms for data flow to achieve multi layered computation for higher accuracy and lower error rate. TensorFlow was released as open source in November 2015 by Google and this move has motivated researchers and scientists to work on this powerful library.

SNMP-MIB data-set was used to train and test the model. the MIB data-set contains 4998 records with 34 attributes (features). The data-set records six different types of Dos attacks and one brute force attack. As shown in below snapshot, ANN has improved detection process and increase the accuracy rate using deep learning with multi layered neural networks. Through the test performance of the model was focused on two main elements which are high detection rate and low error and it is confirmed the accuracy rate is 99.4%