Multi-layer tree liquid state machine recurrent auto encoder for thyroid detection

The proposed work presents thyroid detection strategy by addressing the various challenges faced when raw data is applied to complex neural network like structure. The authors present a Multi-Layer Tree Liquid State Machine Recurrent Auto encoder for the detection of the thyroid nodules. The tree based architecture prevents the loss of original information from dataset when applied to machine learning models like neural network. Liquid State Machine (LSM) prevents the loss of temporal feature of the data from the dataset. The multi layered architecture of the proposed system helps to classify the thyroid stage accurately. The classification rate of the proposed strategy increased when compared to other techniques where the aspect of dataset is not considered.