A simple denoising approach to exploit multi-fidelity data for machine learning materials properties

HIGHLIGHTS

SUMMARY

With the considerable increase in available data, materials science is undergoing a revolution as attested by the multiplication of data-driven studies in recent years1-5. Various advanced machine_learning techniques have been employed in the predictor model, such as attention13, graph convolution11, embedding14, or dimensionality reduction15. The authors focus on various datasets (four with DFT predictions and one with experimental measurements) available for the band gap (see the sub-section "Data" in Methods). This remark is also important in the framework of the machine_learning training process. Indeed, a basic underlying assumption of such approaches . . .

 



If you want to have access to all the content you need to log in!

Thanks :)

Username or Email

Password

Remember me

Lost your password?

If you don't have an account, you can create one here.