A Time-Series Data Augmentation Model through Diffusion and Transformer Integration

IEEE TRANSACTIONS ON HUMAN-MACHINE SYSTEMS 1 A Time-Series Data Augmentation Model through Diffusion and Transformer Integration Y uren Zhang ID, Zhongnan Pu ID, Lei jing ID Member,IEEE Abstract --With the development of Artificial Intelligence, numerous real-world tasks have been accomplished using technology integrated with deep learning. T o achieve optimal performance, deep neural networks typically require large volumes of data for training. Although advances in data augmentation have facilitated the acquisition of vast datasets, most of this data is concentrated in domains like images and speech. However, there has been relatively less focus on augmenting time-series data. T o address this gap and generate a substantial amount of time-series data, we propose a simple and effective method that combines the Diffusion and Transformer models. By utilizing an adjusted diffusion denoising model to generate a large volume of initial time-step action data, followed by employing a Transformer model to predict subsequent actions, and incorporating a weighted loss function to achieve convergence, the method demonstrates its effectiveness. Using the performance improvement of the model after applying augmented data as a benchmark, and comparing the results with those obtained without data augmentation or using traditional data augmentation methods, this approach shows its capability to produce high-quality augmented data. I NTRODUCTION W ITH the development of artificial intelligence (AI), numerous tasks in the real world have been accomplished through technologies combined with deep learning. Typically, a neural network that exhibits excellent performance requires a substantial amount of data for training. V arious types of multi-modal data, such as images, speech, and audio, can now be easily obtained from the Internet. The acquisition of these types of data is no longer an issue. However, due to privacy concerns, costs, and other factors, not all types of data can reach the scale of image or other types data. For instance, the data scale of rare diseases often remains relatively small [1], [2].