Deep recurrent Gaussian Nesterovs recommendation using multi-agent in social networks

Area of Science:

• Artificial Intelligence
• Machine Learning
• Data Science

Background:

• The proliferation of big data in social networks hinders users' ability to extract valuable information.
• Existing recommendation systems struggle with the complexity and scale of real-world data.
• Multi-agent deep learning shows promise but precise recommendations in big data remain a challenge.

Purpose of the Study:

• To propose a novel recommendation system, Deep Recurrent Gaussian Nesterov's Optimal Gradient (DR-GNOG), for optimal and precise recommendations.
• To address the challenge of information overload in social networks through intelligent data processing.
• To improve the accuracy, speed, and recall rate of recommendation systems.

Main Methods:

• The DR-GNOG system employs a multi-layer architecture combining deep learning with a multi-agent approach.
• A Tweet Accumulator Agent feeds user tweets into the input layer.
• The first hidden layer uses Gaussian Nesterov's Optimal Gradient for optimized tweet classification.
• The second hidden layer features a Deep Recurrent Predictive Recommendation model to mitigate vanishing gradient issues.
• A hyperbolic activation function is utilized in the output layer for predictive recommendation.

Main Results:

• The DR-GNOG method demonstrated significant improvements over existing GANCF and Bootstrapping methods.
• Recommendation accuracy was enhanced by 13-21%.
• Recommendation time was improved by 22-32%.
• Recall rate saw an increase of 15-22%.

Conclusions:

• The proposed DR-GNOG system offers a superior solution for precise and efficient recommendations in big data environments.
• The integration of multi-agent deep learning and advanced gradient optimization techniques effectively addresses key challenges in social network data.
• Experimental results confirm the substantial performance gains of DR-GNOG in accuracy, speed, and recall.