The field of sign language recognition and geophysical inversions is experiencing significant advancements with the application of transformer-based methods. Recent developments have shown that these methods can effectively capture spatial and temporal dependencies in sign language data, leading to improved recognition accuracy. Additionally, transformer-based approaches have been successfully applied to geophysical inversions, enabling more accurate and efficient recovery of subsurface physical property models. Notably, the use of decision feedback mechanisms and semi-supervised learning techniques has improved the performance of these methods in scenarios with limited pilot data. Furthermore, the integration of kinematic information and motion gesture primitives has enhanced the realism and accuracy of sign language recognition systems. Overall, these innovations have the potential to improve communication between deaf and non-deaf individuals and advance our understanding of subsurface structures. Noteworthy papers include: Transformer-based Wireless Symbol Detection Over Fading Channels, which proposes a novel decision feedback mechanism for improved symbol detection. A multitask transformer to sign language translation using motion gesture primitives, which introduces a gloss learning representation for more suitable translation. Siformer: Feature-isolated Transformer for Efficient Skeleton-based Sign Language Recognition, which presents a kinematic hand pose rectification method for enforcing constraints. Interpretable Deep Learning Paradigm for Airborne Transient Electromagnetic Inversion, which proposes a unified and interpretable deep learning inversion paradigm based on disentangled representation learning. DREMnet: An Interpretable Denoising Framework for Semi-Airborne Transient Electromagnetic Signal, which proposes an interpretable decoupled representation learning framework for robust and interpretable denoising.