Software Vulnerability Assessment and Risk Identification Using Recurrent Nural Network

Abstract

Identifying software vulnerabilities is an effective method for improving technological quality and optimizing testing management by enabling the early detection of deficiencies in simulation models before the actual testing phase begins. These predictive insights assist technology developers in efficiently allocating resources to the components most susceptible to flaws. This study introduces a software vulnerability prediction model utilizing a Deep Learning (DL) approach. A Recurrent Neural Network (RNN) is employed to classify source code, incorporating various soft computing techniques. Several preprocessing and data filtration methods have been applied to ensure data normalization and balance. To create a Vector Space Model (VSM), Term Frequency-Inverse Document Frequency (TF-IDF) and relationship-based feature extraction techniques have been implemented. The classification process is conducted using RNN on both training and validation datasets. The evaluation of performance of proposed work is performed using various real-time and synthetic accessible databases. It is observed from the experimental results that the proposed framework performs better when different datasets are evaluated.