Research on multimedia content compression and transmission algorithm based on deep learning

This paper focuses on the research of multimedia content compression and transmission algorithm based on deep learning, aiming to solve the shortcomings of traditional algorithms in compression ratio, transmission rate and reconstruction quality. A deep learning algorithm integrating attention mechanism and generative adversarial network is innovatively proposed. In the experimental stage, a hardware environment consisting of NVIDIA Tesla V100 GPU and Intel Xeon Platinum 8280 CPU was built, and the research was carried out based on the PyTorch framework. Public datasets such as COCO were used and divided into training set, validation set and test set in a ratio of 8:1:1. Comparative experiments show that the new algorithm has an average compression ratio of 30:1 in image compression, which is 50% higher than JPEG; in a network environment with 100Mbps bandwidth and 50ms delay, the transmission rate reaches 80Mbps, which is 30% higher than TCP; the PSNR of the reconstructed image reaches 35dB and the SSIM reaches 0.92. The experimental results fully verify the significant improvement of the new algorithm in multimedia content compression and transmission performance, and provide a better solution for multimedia data processing. Keywords: deep learning; multimedia compression; transmission algorithm; attention mechanism; generative adversarial network