The field of data compression is witnessing significant developments, particularly in non-linear transformations and their applications. Researchers are exploring the interplay between information and computation in non-linear transform-based compression, highlighting key trade-offs between compact representations and computational cost. These advancements have far-reaching implications for various tasks, including classification, denoising, and generative AI. Noteworthy papers include:

• One that explores relative entropy coding, a mathematical framework for efficient communication of uncertain information, and its integration with modern machine learning pipelines.
• Another that investigates the exploration-exploitation tradeoff in universal lossy compression, deriving robust cost-directed algorithms for sequential lossy compression.
• A paper on communication-aware map compression for online path-planning, which formulates the compression design problem as a rate-distortion optimization problem, enabling efficient and real-time implementation.