Quantum Diffusion Models for Few-Shot Learning

TL;DR

Three new frameworks employing quantum diffusion model (QDM) as a solution for the few-shot learning are proposed: label-guided generation inference (LGGI); label-guided denoising inference (LGDI); and label-guided noise addition inference (LGNAI).

Abstract

Modern quantum machine learning (QML) methods involve the variational optimization of parameterized quantum circuits on training datasets, followed by predictions on testing datasets. Most state-of-the-art QML algorithms currently lack practical advantages due to their limited learning capabilities, especially in few-shot learning tasks. In this work, we propose three new frameworks employing quantum diffusion model (QDM) as a solution for the few-shot learning: label-guided generation inference (LGGI); label-guided denoising inference (LGDI); and label-guided noise addition inference (LGNAI). Experimental results demonstrate that our proposed algorithms significantly outperform existing methods.

Figures (15)

• Figure 1: Various types of variational quantum circuits (VQC).
• Figure 2: Generated images using QDDM under the guidance of different labels. The input to the model is random noise.
• Figure 3: Framework of QDDM. Noise Predictor is employed to estimate the noise present in the noisy image data.
• Figure 4: Framework of QDDM-based Label-Guided Generation Inference (QDiff-LGGI). The gray-filled circle represents the embedded label.
• Figure 5: Framework of QDDM-based Label-Guided Noise Addition Inference (QDiff-LGNAI). The term $\hat{\epsilon}_m^{n}$ represents the predicted noise at step $m$ associated with label $n$. $L_0$/$L_1$-loss denotes the difference between the true noise and the predicted noise under the guidance of different labels $L_i$.