Organized around 10 modules, this program helps participants broaden and deepen their Python programming skills for machine learning applications. This technical knowledge can be applied to any industry integrating machine learning and artificial intelligence into their digital drivers.
As you begin, you will learn to use a decision tree to make predictions and, given labeled training examples, you will learn a decision tree.
Binary Logistic Regression
Given i.i.d. data and parameters of a logistic regression distribution, you will learn to compute conditional likelihood and learn to implement stochastic gradient descent for binary logistic regression.
In machine learning, there are fundamental algorithms. In this module, you will learn to use the k-NN algorithm to classify points given a simple dataset and implement a full decision tree for learning and prediction.
As you discover ways to combat overfitting, you will convert a nonlinear dataset to a linear dataset in higher dimensions, manipulate the hyperparameters of L1 and L2 regularization implementations and identify the effects on magnitude and sparsity of parameters.
Building your skills in Python, you will employ model selection techniques to select k for the k-NN algorithm and implement a grid search to select multiple hyperparameters for a model.
Combine simpler models as components to build up feed-forward neural network architectures and write mathematical expressions in scalar form defining a feed-forward neural network.
Creating machine learning solutions can require refinement of the inner workings of algorithms, including adapting the k-NN algorithm for classification to regression, adapting decision trees for classification to regression, as well as implementing learning for linear regression using gradient descent.
Adding to your deep knowledge of algorithmic applications, you will learn to carry out the backpropagation algorithm on a simple computation graph over scalars and instantiate the backpropagation algorithm for a neural network.
In this module, you will determine how convexity affects optimization and implement linear regression with optimization by stochastic gradient descent.
K-Means and Others Learning Paradigms
In addition to exploring solutions to practical challenges in this final module, you will learn to implement the k-means algorithm and recognize and explain challenges in selecting the number of clusters.