In this project, three different tasks were included: "Object recognition", "Image quantization", and "The impact of dimensionality reduction in classification". "Object recognition" was an unsupervised sub-problem of the generic object recognition task. We were provided with "The coins dataset", which is a scikit-learn toy dataset. It includes a single picture that displays 24 Greek coins from Pompeii. The task was to identify the coins by using the K-Means clustering algorithm and to experiment with different values of hyper-parameters, especially the initial number of clusters. In "Image quantization" task, we had to work with the flower.jpg image from load_sample_image of sklearn.datasets. We had to quantize the input image by applying k-Means clustering and perform experiments with different numbers of initial clusters. Finally, in "The impact of dimensionality reduction in classification" task we had to work with a high-dimensional text dataset of scikit-learn, called "The 20newsgroup". We had to apply PCA for various sizes of the input space (e.g. 50, 100, 500, 1000, 10000, and so on), as well as to compare the performance of LogisticRegression, Random Forest Classifier, and Multilayer Perceptron on both the reduced and original dimensional spaces.