MRAM based neuromorphic cell for Artificial Intelligence.

Applications in artificial intelligence architectures require cell elements that can take multiple states based on different cell inputs, others need to generate random numbers efficiently. Hardware realizations of these can be realized on magnetic non-volatile memory (MRAM) cells. The thesis project investigates how to exploit MRAM Technologies for applications in Artificial Intelligence (AI). Connecting individual MRAM cell in parallel or in series will give an overall multilevel measured resistance that depend on the actual states of individual cells. A Magnetic Tunnel Junction can also be used as a Random Bit Generator. It can be done by finding the voltage at which the switching probability to go from one state to the other is 50%, the transitions will be intrinsically random.