A recent study published on arXiv investigates the impact of different loss functions on robustness and the optimization landscape in non-convex matrix sensing problems, particularly under noisy conditions. The research identifies limitations of the commonly used mean squared error (MSE) loss when dealing with non-Gaussian noise, which can degrade performance. To address this issue, the authors propose a kernel-based loss function designed to enhance robustness against such noise. This kernel optimal loss function aims to improve the stability and reliability of matrix sensing algorithms by better handling noise characteristics that MSE struggles with. The study contributes to ongoing discussions about optimization methods in machine learning by highlighting how loss function choice affects both robustness and the shape of the optimization landscape. These findings align with broader research efforts focused on developing more resilient and effective loss functions for complex data scenarios. Overall, the proposed kernel loss offers a promising alternative to traditional approaches in noisy matrix sensing tasks.