A Streaming model for Generalized Rayleigh with extension to Minimum Noise Fraction

Area of Science:

• Optimization Algorithms
• Big Data Analytics
• Signal Processing

Background:

• Rayleigh quotient optimization involves maximizing a rational function, often a max-min problem.
• Traditional methods struggle with big-data scenarios due to memory limitations.

Purpose of the Study:

• To develop a low-rank, streaming solution for Rayleigh quotient optimization.
• To apply this method for maximizing the signal-to-noise ratio (SNR) in big data.

Main Methods:

• A novel low-rank, streaming approach for Rayleigh quotient optimization.
• Application to big-data signal-to-noise ratio maximization for static and dynamic data.

Main Results:

• The streaming implementation achieved faster processing times than standard in-memory methods.
• Demonstrated trade-offs in accuracy, speed, and storage using synthetic and real data.

Conclusions:

• The proposed low-rank, streaming solution is effective for big-data Rayleigh quotient optimization.
• This approach offers a scalable and efficient alternative for large-scale SNR maximization.