High-Value Patents Recognition with Random Forest and Enhanced Fire Hawk Optimization Algorithm.

Area of Science:

• Intellectual Property Analytics
• Artificial Intelligence
• Optimization Algorithms

Background:

• High-value patents are critical indicators of innovation and new product development.
• Identifying high-value patents is challenging due to skewed distributions and limitations of current machine learning methods.
• Existing optimization algorithms struggle with slow convergence and local optima on complex patent data.

Purpose of the Study:

• To develop an advanced optimization algorithm for effective hyperparameter tuning in patent analysis.
• To address the limitations of existing methods in identifying high-value patents.
• To enhance the accuracy and stability of machine learning models for patent recognition.

Main Methods:

• Proposal of the Enhanced Fire Hawk Optimizer (EFHO) incorporating adaptive tent chaotic mapping, hunting prey, inertial weight, and enhanced flee strategies.
• Application of EFHO to optimize Random Forest hyperparameters for high-value patent recognition.
• Evaluation of EFHO's performance on benchmark tests and real-world patent datasets.

Main Results:

• EFHO demonstrated superior convergence speed, accuracy, and robustness compared to standard optimization benchmarks.
• The EFHO-optimized Random Forest model achieved higher accuracy and classification stability than other methods for high-value patent recognition.
• EFHO effectively overcame common issues like slow convergence and local optima trapping.

Conclusions:

• The Enhanced Fire Hawk Optimizer (EFHO) offers a robust and efficient solution for hyperparameter tuning in machine learning.
• EFHO shows significant practical value in accurately identifying high-value patents from large datasets.
• This research contributes to overcoming challenges in patent analysis and machine learning optimization.