A system processes information in discrete steps and updates its internal memory at each step. The update is performed by a neural network that follows the general rule: New_Memory = Update_Function(Current_State, Previous_Memory). If the system is currently processing Step 10, what specific inputs are required by the Update_Function to calculate the memory for this step?

Analysis of Memory Update Inputs

Diagnosing a Faulty Memory Update Mechanism

RNN-like Configuration for Neural Memory