Explore how to how to build and train the two most versatile and ubiquitous types of deep learning models in TensorFlow.
Learning Objectives
TensorFlow: Simple Regression and Classification Models
- Course Overview
- recognize linear regression problems and extend that to general machine learning problems
- recognize how model parameter training happens via gradient descent to find minimum loss
- load a dataset and explore its features and labels
- choose the right form of data to feed into the linear regression model
- build a base model for comparison with scikit-learn
- create placeholders, training variables, and instantiate optimizers to use with regression
- train model parameters using a session and the training dataset, and visualize the result with Matplotlib
- demonstrate how to interpret the loss and summaries on TensorBoard
- choose the high-level Estimator API for common use cases
- train a regression model using the high-level Estimator API
- evaluate and predict housing prices using estimators
- identify classification problems and recall logistic regression for classification
- recognize cross entropy as the loss function for classification problems and use softmax for n-category classification
- identify data as being a continuous range or comprised of categorical values
- work with training and test data to predict heart disease
- train the high-level estimator for classification and use it for prediction
- describe basic concepts of the linear regression machine learning model
- describe basic concepts of the binary classification machine learning model
Framework Connections
Specialty Areas
- Software Development
Feedback
If you would like to provide feedback for this course, please e-mail the NICCS SO at NICCS@hq.dhs.gov.