Quantum Train: A Novel Hybrid Quantum-Classical Machine Learning Approach in the Model Compression Perspective

  • Event Date: 2024-09-16
  • Quantum information and communication
  • Speaker: Prof. Hsi-Sheng Goan (Dept. of Physics, NTU)  /  Host: Prof. Yueh-Nan Chen(NCKU)
    Place: R36173, 1F, Dept. of Physics, Building of Science College, NCKU

Time: 12:10, Monday, September 16, 2024
Speaker: Prof. Hsi-Sheng Goan
               Dept. of Physics, NTU
Title: Quantum Train: A Novel Hybrid Quantum-Classical Machine Learning Approach in the Model Compression Perspective
Place: R36173, 1F, Dept. of Physics, Building of Science College, NCKU

Abstract:
We introduce the quantum-train (QT) framework, a novel approach that integrates quantum computing with classical machine learning algorithms to address significant challenges in data encoding, model compression, and inference hardware requirements. Even with a slight decrease in accuracy, QT achieves remarkable results by employing a quantum neural network alongside a classical mapping model, which significantly reduces the parameter count from M to O(polylog(M)) during training. Our experiments demonstrate QT’s effectiveness in classification tasks, offering insights into its potential to revolutionize machine learning by leveraging quantum computational advantages. This approach not only improves model efficiency but also reduces generalization errors, showcasing QT’s potential across various machine learning applications.

1/0