Rotary Masked Autoencoders are Versatile Learners
Provides a versatile, architecture-agnostic method for irregular time-series representation learning, reducing the need for task-specific designs.
RoMAE extends Masked Autoencoders with Rotary Positional Embedding to handle irregular time-series without architectural specializations, surpassing specialized models on the DESC ELAsTiCC Challenge while maintaining performance on images and audio.
Applying Transformers to irregular time-series typically requires specializations to their baseline architecture, which can result in additional computational overhead and increased method complexity. We present the Rotary Masked Autoencoder (RoMAE), which utilizes the popular Rotary Positional Embedding (RoPE) method for continuous positions. RoMAE is an extension to the Masked Autoencoder (MAE) that enables interpolation and representation learning with multidimensional continuous positional information while avoiding any time-series-specific architectural specializations. We showcase RoMAE's performance on a variety of modalities including irregular and multivariate time-series, images, and audio, demonstrating that RoMAE surpasses specialized time-series architectures on difficult datasets such as the DESC ELAsTiCC Challenge while maintaining MAE's usual performance across other modalities. In addition, we investigate RoMAE's ability to reconstruct the embedded continuous positions, demonstrating that including learned embeddings in the input sequence breaks RoPE's relative position property.