Peak Signal-to-Noise Ratio (PSNR)

Peak signal-to-noise ratio (PSNR) is an engineering term for the ratio between the maximum possible power of a signal and the power of corrupting noise that affects the fidelity of its representation. Because many signals have a very wide dynamic range, PSNR is usually expressed as a logarithmic quantity using the decibel scale. PSNR is commonly used to quantify reconstruction quality for images and video subject to lossy compression.

PSNR can be indirectly linked to the Robustness objective, as it quantifies how well an AI system preserves signal quality in the presence of noise or distortions. High PSNR values suggest that the system maintains reliable performance under certain adverse conditions (e.g., noisy inputs), which is a component of robustness. However, this connection is limited to technical robustness in image quality and does not extend to broader system reliability or resilience.

Trustworthy AI Relevance

This metric addresses Robustness and Transparency by quantifying relevant system properties. Robustness: PSNR quantifies how much an image/output deviates from a reference in terms of pixel-wise error; it is directly relevant to assessing noise resilience, compression / reconstruction fidelity, and the model’s consistency under input degradation or distribution shifts. Monitoring PSNR across conditions can reveal drops in performance under adverse/noisy inputs and help evaluate and improve robustness.

Ragas Documentation: Response Relevancy