Abstract:The edge intelligent computing technology can reduce the delay and energy consumption of deep learning model reasoning through the collaborative terminal acquisition equipment and edge server. We apply the neural network to the edge computing and build a data augmentation computing model based on the sparse data volume. Then, we get an intelligent generative image after the network training to achieve the effect of enhancement computing. In this paper, we choose a relatively small number of national elements and the generative adversarial network (GAN) as the experimental data calculation set and network model. First, we normalize the preprocessing of the collected data to form the initial sample data set. Second, the model extracts the feature vector by input image to the convolution neural network (CNN) layer. After that, we use a random noise vector z which follows a Gaussian distribution as the initial input of the conditional generative adversarial network (CGAN). The feature vector extracted from the image is taken as a label and a condition constraint of the CGAN to train the parameters of the CGAN. Finally, the trained CGAN model is used to complete the data augmentation computing. A total of 350 samples were collected, and 97 sample images were actually applied for data augmentation. The enhanced data set of this model is as many as 1,700 samples, and it is found that the generated image is of good quality by using this data set for the peak signal-to-noise ratio (PSNR) detection, which is of innovative value in the standard of real samples.

Scaling-based Data Augmentation for Generative Models and its Theoretical Extension

Toward Understanding Generative Data Augmentation

Generative Expansion of Small Datasets: An Expansive Graph Approach

Differentiable Augmentation for Data-Efficient GAN Training

Augmenting data with generative adversarial networks: An overview

Diffusion-GAN: Training GANs with Diffusion

Restrained Generative Adversarial Network against Overfitting in Numeric Data Augmentation

On Data Augmentation for GAN Training

DifAugGAN: A Practical Diffusion-style Data Augmentation for GAN-based Single Image Super-resolution

On Stabilizing Generative Adversarial Training with Noise

Data Augmentation Computing Model Based on Generative Adversarial Network

Learning from Bad Data Via Generation

Effective Data Augmentation with Multi-Domain Learning GANs

MS^3D: A RG Flow-Based Regularization for GAN Training with Limited Data

LatentAugment: Data Augmentation via Guided Manipulation of GAN's Latent Space

Stabilizing Training of Generative Adversarial Networks through Regularization

Performance Study of Image Data Augmentation by Generative Adversarial Networks

MS$^3$D: A RG Flow-Based Regularization for GAN Training with Limited Data

ScoreMix: A Scalable Augmentation Strategy for Training GANs with Limited Data

Data Augmentation for Image Classification using Generative AI

An information theoretic limit to data amplification