(a) A data scientist wants to build a Deep Learning model using a single LSTM cell for forecasting a time-series. The training dataset has several records with 15 time steps each. Each time step consists of three-feature normalized numeric data. If total learnable parameters in this model are 180, what is the dimensionality of the short state? [2] (b) A healthcare startup wants to analyze MRI scans to detect tumors automatically. Each MRI scan is a sequence of 50 grayscale images of size 256 x 256. Describe the neural network architecture you would recommend for this task. Justify your choice. [2] (e) You are building a model for detecting fake news articles. Describe the neural network architecture you would recommend for this task. Justify your choice. [2] (d) A tech startup aims to build a chatbot that can understand the context of a conver- sation, remember past interactions, and respond appropriately. Describe the neural network architecture you would recommend for this task. Justify your choice. Discuss how you would handle varying lengths of input sequences in the model. [2]