Abstract:More and more attention has been paid to computer security, and its vulnerabilities urgently need more sensitive solutions. Due to the incomplete data of most vulnerability libraries, it is difficult to obtain pre-permission and post-permission of vulnerabilities, and construct vulnerability exploitation chains, so it cannot to respond to vulnerabilities in time. Therefore, a vulnerability extraction and prediction method based on improved information gain algorithm is proposed. Considering the accuracy and response speed of deep neural network, deep neural network is adopted as the basic framework. The Dropout method effectively reduces overfitting in the case of incomplete data, thus improving the ability to extract and predict vulnerabilities. These experiments confirmed that the excellent F1 and Recall of the improved method reached 0.972 and 0.968, respectively. Compared to the function fingerprints vulnerability detection method and K-nearest neighbor algorithm, the convergence is better. Its response time is 0.12 seconds, which is excellent. To ensure the reliability and validity of the proposed method in the face of missing data, the reliability and validity of Mask test are verified. The false negative rate was 0.3% and the false positive rate was 0.6%. The prediction accuracy of this method for existing permissions reached 97.9%, and it can adapt to the development of permissions more actively, so as to deal with practical challenges. In this way, companies can detect and discover vulnerabilities earlier. In security repair, this method can effectively improve the repair speed and reduce the response time. The prediction accuracy of post-existence permission reaches 96.8%, indicating that this method can significantly improve the speed and efficiency of vulnerability response, and strengthen the understanding and construction of vulnerability exploitation chain. The prediction of the posterior permission can reduce the attack surface of the vulnerability, thus reducing the risk of breach, speeding up the detection of the vulnerability, and ensuring the timely implementation of security measures. This model can be applied to public network security and application security scenarios in the field of computer security, as well as personal computer security and enterprise cloud server security. In addition, the model can also be used to analyze attack paths and security gaps after security accidents. However, the prediction of post-permissions is susceptible to dynamic environments and relies heavily on the updated guidance of security policy rules. This method can improve the accuracy of vulnerability extraction and prediction, quickly identify and respond to security vulnerabilities, shorten the window period of vulnerability exploitation, effectively reduce security risks, and improve the overall network security defense capability. Through the application of this model, the occurrence frequency of security vulnerability time is reduced effectively, and the repair time of vulnerability is shortened.

An approach for predicting multiple-type overflow vulnerabilities based on combination features and a time series neural network algorithm

Ignnvd: A Novel Software Vulnerability Detection Model Based on Integrated Graph Neural Networks

Combining Software Metrics and Text Features for Vulnerable File Prediction

Vulnerability Time Series Prediction Based on Multivariable LSTM

Multi-class vulnerability prediction using value flow and graph neural networks

Predicting Missing Information of Key Aspects in Vulnerability Reports

BovdGFE: Buffer Overflow Vulnerability Detection Based on Graph Feature Extraction

Vulnerability extraction and prediction method based on improved information gain algorithm

Hybrid semantics-based vulnerability detection incorporating a Temporal Convolutional Network and Self-attention Mechanism

Vulnerability Detection via Multiple-Graph-Based Code Representation

Vulnerability Severity Prediction Model for Software Based on Markov Chain.

VulMPFF: A Vulnerability Detection Method for Fusing Code Features in Multiple Perspectives

Predicting flood susceptibility using LSTM neural networks

Vulnerability Detection in C/C++ Code with Deep Learning

Vulnerability detection based on transformer and high‐quality number embedding

A software vulnerability detection method based on deep learning with complex network analysis and subgraph partition

Progress in Research on Buffer Overflow Vulnerability Analysis Technologies

An Improved Vulnerability Exploitation Prediction Model with Novel Cost Function and Custom Trained Word Vector Embedding

FastEmbed: Predicting vulnerability exploitation possibility based on ensemble machine learning algorithm

Software vulnerable functions discovery based on code composite feature

Predicting Severity of Software Vulnerability Based on Grey System Theory.