Beyond Classical: A Quantum Approach to Time-Series Prediction

A quantum computing approach for time-series prediction is proposed, starting with an overview of classical models like LSTMs and their limitations. It then introduces the fundamentals of Quantum Machine Learning (QML), focusing on Parameterized Quantum Circuits as building blocks for hybrid models. The primary focus is a novel architecture called Coherent Quantum Long Short-Term Memory (CoQS-LSTM), which integrates quantum components into the traditional LSTM cell structure. This hybrid model aims to offer a more efficient way to learn from temporal data by leveraging quantum computation. The discussion includes benchmarks and results to evaluate the capabilities and limitations of this new quantum-enhanced recurrent network.