| Name: | Chandan Ganjigere Nagarajappa |
| Affiliation: | University of KwaZulu-Natal |
| Conference ID : | ASI2022_482 |
| Title : | Using Machine learning to classify CMB temperature maps |
| Authors : | Mr. Chandan Ganjigere Nagarajappa Prof. Yin-Zhe Ma Dr. Cheng Cheng |
| Abstract Type: | Poster* |
| Abstract Category : | General Relativity and Cosmology |
| Abstract : | Extracting fNL using two-point statistics is computationally challenging, to overcome this challenge it has been effective to use the machine learning algorithms. The use of machine learning techniques like Convolutional Neural Network (CNN) and Graph Neural Network (GNN) bypasses the immense computational power and time required to detect the non-Gaussianity in the CMB temperature maps. In this work, the machine learning techniques are applied to classify the CMB maps containing non-Gaussianity from the CMB maps which are purely Gaussian. This is achieved by generating the Gaussian and non-Gaussian CMB temperature maps by varying fNL values from the lower limit of -50 to upper limit of +50. The generated CMB temperature maps are fed to the machine which is trained to identify the non-Gaussianity in the maps and further classify the CMB temperature maps into Pure Gaussian maps or non-Gaussian maps. The work further attempts to extract the fNL parameter from the generated CMB temperature maps. |