In this project we implemented different MCMC techniques and applied them to image denoising by assuming an underlying Ising model and to finding candidate solutions for the Travelling Salesman Problem (TSP).
Implementation of the DA algorithm according to Deterministic annealing for clustering, compression, classification, regression, and related optimization problems and empirical analysis of its phase transition behaviour. Reference paper at: https://ieeexplore.ieee.org/document/726788
We implemented Histogram Clustering for Image Segmentation, both via maximum a posterior probability (MAP) and deterministic annealing (DA) for predicting the cluster membership for each pixel. Reference paper at: https://ieeexplore.ieee.org/document/784981
We implemented Constant Shift Embedding (CSE), a method in which we embed pairwise clustering problems (where data points may lie in a discrete structure, such as a graph) in vector spaces in order to render classical clustering techniques in vector spaces (e.g. K-Means) available. You can find the reference paper at: https://ieeexplore.ieee.org/document/1251147