Prediction in Autism by Partial Differential Diffusion and Frobenius Convolutional Light Gated Recurrent in Toddlers

World is discerned emergence of innumerable sicknesses which cannot perceived medically, identified, like Autism Spectrum Disorder (ASD). It takes holds of numerous behavioral spheres, as well as social and linguistic expertise and conventionalized actions. This ASD disability is a significant neuro-developmental disorder. Owing to the reason that several additional mental sicknesses is spectacularly homogeneous indicators to individuals of ASD, diagnosing ASD can laborious as well as time consuming. Also premature ASD diagnosis depend on distinct physiological features appears realistic through increasing employing of DL based methods in forecasting numerous human diseases. A novel ASD prediction method called, Partial Differential Diffusion and Frobenius Convolutional Light Gated Recurrent (PDD-FCLGR) in toddlers is introduced. PDD-FCLGR technique is divide as preprocessing and ASD prediction. The ASD toddler data experiment is subjected to preprocessing using Partial Differential Diffusion-based Preprocessing model to obtain noise reduced and dimensionality reduced data. Here, partial differential function is applied to remove noise whereas Discrete Gaussian Diffusion map is applied to produce dimensionality reduced data. Second, the Frobenius Convolutional Light Gated Recurrent Unit (FCLiGRU) ASD prediction model is designed by combining Frobenius Convolutional Neural Network and Light Gated Recurrent Unit. Extensive simulation is proposed to be done using Python high level general purpose programming language and detailed comparison is proposed to be made with conventional methods in terms of RMSE, ASD prediction accuracy, ASD prediction time and specificity.