Ramin Barati

project owner

Unsupervised Learning

Our team proposes the implementation of clustering heuristics in MeTTa, focusing initially on K-Means, Hierarchical Clustering, Spectral Clustering, and the Gaussian Mixture Model (GMM). Leveraging our experience in variational inference and EM algorithms, including past work on a generalized GMM model, we aim to align our approach with MeTTa’s design philosophy, collaborating closely with the Hyperon community to ensure an intuitive fit. By adhering to conventions from libraries like numpy, scikit-learn, and matplotlib, our project will create a foundational toolkit for unsupervised learning in MeTTa, promoting broader adoption and setting the stage for more advanced applications.