The field of machine learning is witnessing significant developments in representation learning and novel methodologies. Researchers are exploring new principles and frameworks to achieve identifiable disentangled representations, which is crucial for understanding the underlying factors of variation in observed data. Additionally, there is a growing interest in applying concepts from signal processing and formal language theory to improve classification and feature selection tasks. The use of alternative metrics, such as p-adic metrics, is also being investigated to better capture hierarchical relationships in data. Noteworthy papers in this area include: Mechanistic Independence: A Principle for Identifiable Disentangled Representations, which introduces a unified framework for disentanglement through mechanistic independence. A signal separation view of classification proposes an alternative approach for classification using localized trigonometric polynomial kernels. Linear Regression in p-adic metric spaces presents a theoretical foundation for machine learning in p-adic metric spaces, which naturally respect hierarchical structure.