Optimisation Challenge for a Superconducting Adiabatic Neural Network That Implements XOR and OR Boolean Functions

Area of Science:

• * Condensed matter physics
• * Quantum computing
• * Artificial intelligence

Background:

• * Artificial neural networks (ANNs) are computational models inspired by biological neural networks.
• * Josephson junctions are key components in superconducting electronics, enabling unique quantum phenomena.
• * Adiabatic quantum computing utilizes quantum fluctuations to find solutions to complex problems.

Purpose of the Study:

• * To design simple analog artificial neural networks using adiabatic Josephson cells.
• * To develop a gradient descent method for optimizing circuit parameters.
• * To demonstrate the network's functionality with XOR and OR logical operations.

Main Methods:

• * Design of analog ANNs utilizing adiabatic Josephson cells with a sigmoid activation function.
• * Application of a gradient descent method for parameter adjustment.
• * Implementation and testing of XOR and OR logic gate functionalities.

Main Results:

• * Successful design of analog ANNs based on adiabatic Josephson cells.
• * Efficient signal transmission between network layers achieved through parameter optimization.
• * Demonstrated functionality for implementing XOR and OR logical operations.

Conclusions:

• * Adiabatic Josephson cells offer a viable platform for analog ANNs.
• * Gradient descent provides an effective method for parameter tuning in these networks.
• * The proposed design is suitable for implementing basic logic functions.