20230049479. COMPUTER-IMPLEMENTED METHOD FOR ACCELERATING CONVERGENCE IN THE TRAINING OF GENERATIVE ADVERSARIAL NETWORKS (GAN) TO GENERATE SYNTHETIC NETWORK TRAFFIC, AND COMPUTER PROGRAMS OF SAME simplified abstract (Telefonica, S.A.)

COMPUTER-IMPLEMENTED METHOD FOR ACCELERATING CONVERGENCE IN THE TRAINING OF GENERATIVE ADVERSARIAL NETWORKS (GAN) TO GENERATE SYNTHETIC NETWORK TRAFFIC, AND COMPUTER PROGRAMS OF SAME

Organization Name

Telefonica, S.A.

Inventor(s)

Alberto Mozo Velasco of Madrid (ES)

Sandra Gomez Canaval of Madrid (ES)

Antonio Pastor Perales of Madrid (ES)

Diego R. Lopez of Madrid (ES)

Edgar Talavera Munoz of Madrid (ES)

COMPUTER-IMPLEMENTED METHOD FOR ACCELERATING CONVERGENCE IN THE TRAINING OF GENERATIVE ADVERSARIAL NETWORKS (GAN) TO GENERATE SYNTHETIC NETWORK TRAFFIC, AND COMPUTER PROGRAMS OF SAME - A simplified explanation of the abstract

This abstract first appeared for US patent application 20230049479 titled 'COMPUTER-IMPLEMENTED METHOD FOR ACCELERATING CONVERGENCE IN THE TRAINING OF GENERATIVE ADVERSARIAL NETWORKS (GAN) TO GENERATE SYNTHETIC NETWORK TRAFFIC, AND COMPUTER PROGRAMS OF SAME

Simplified Explanation

The patent application proposes a method for speeding up the training of generative adversarial networks (GANs) to generate synthetic network traffic. This method ensures that the training converges in a shorter time period compared to existing GAN networks. The innovation allows for the creation of simulated network traffic data with similar characteristics to real datasets, without using any part of the real dataset. It also enables the generation of diverse types of data such as IP traffic and network attacks. Additionally, the method can detect changes in network traffic patterns.

• The method accelerates convergence in GAN training for generating synthetic network traffic.
• It allows for the creation of simulated data with characteristics similar to real network traffic datasets.
• The method can generate diverse types of data, including IP traffic and network attacks.
• It enables the detection of changes in network traffic patterns.

Potential Applications

This technology has potential applications in various areas related to network traffic analysis and simulation, including:

• Network security: The generated synthetic network traffic can be used to train and test intrusion detection systems, allowing for more robust security measures.
• Network performance testing: Simulated network traffic can be used to evaluate the performance and scalability of network infrastructure and applications.
• Anomaly detection: By comparing the generated synthetic traffic with real traffic, anomalies in network behavior can be identified.
• Training data augmentation: The synthetic data can be used to augment real datasets, providing a larger and more diverse training set for machine learning models.

Problems Solved

The proposed method addresses several problems in the training of GANs for generating synthetic network traffic:

• Convergence time: The method accelerates the convergence of GAN training, reducing the time required to generate realistic synthetic data.
• Data privacy: By generating synthetic data without using any part of real datasets, privacy concerns associated with using sensitive network traffic data are mitigated.
• Data diversity: The method allows for the creation of diverse types of network traffic data, enabling more comprehensive analysis and testing.

Benefits

The use of this technology offers several benefits:

• Time savings: The accelerated convergence of GAN training reduces the time required to generate synthetic network traffic.
• Data privacy protection: By not using any real dataset, the method ensures the privacy of sensitive network traffic data.
• Enhanced analysis capabilities: The ability to generate diverse types of network traffic data allows for more comprehensive analysis and testing.
• Improved network security: The synthetic data can be used to train and test intrusion detection systems, leading to more effective network security measures.

Original Abstract Submitted

proposed are a computer-implemented method for accelerating convergence in the training of generative adversarial networks (gan) to generate synthetic network traffic, and computer programs of same. the method allows the gan network to ensure that the training converges in a limited time period less than the standard training period of existing gan networks. the method allows results to be obtained in different use scenarios related to the generation and processing of network traffic data according to objectives such as the creations of arbitrary amounts of simulated data (a) with characteristics (statistics) similar to real datasets obtained from real network traffic, but (b) without including any part of any real dataset; diversity in the type of data to be created: ip traffic, network attacks, etc.; and the detection of changes in the network traffic patterns analysed and generated.