Tag: scikit-learn

• Unsupervised Learning of Gaussian Mixture Models With the EM Algorithm (13 Dec 2019)
  The Expectation-Maximization algorithm is a powerful iterative method for calculating maximum likelihood estimates from unlabeled data. In this article, we will be exploring its use in Gaussian mixture models to perform the task of clustering with the Scikit-learn digits dataset. The performance of the EM-algorithm is then compared to a vanilla k-Means implementation provided in Scikit-learn.
• Comparing the Performance of Various Classifiers on the Scikit-learn Digits and MNIST Datasets (06 Dec 2019)
  In this article we will be implementing several classifiers to predict the class labels of the Scikit-learn digits and MNIST_Light datasets.
• Decision Tree Learning with CART and ID3 Algorithms (15 Nov 2019)
  Often times, classification can be framed as a sort of questioning-answering system. Questions are asked about the input data which aid the model in determining a prediction. One example of a 'question' that a model might ask is 'does this input image contain this attribute value?' Decision trees naturally help to structure this kind of if-then hierarchical decision-making by defining a series of questions that lead to a class label or value. In this article, we will explore several algorithms for constructing the two types of decision trees; the ID3 algorithm for Classification Trees and the CART for Regression Trees. While reading along, you will be able to implement your own ID3 algorithm from scratch using the code provided in this notebook.