Classif.ai
An interactive web-based tutorial and application that leverages data visualization and learning-by-doing to instruct novices about Machine Learning.
Project Scope
05-839 Interactive Data Science
04.2017-05.2017
Advisors: Jennifer Mankoff, Nikola Banovic
Team: Shouvik Mani, Guanqiao Yin, Jamie Diner, Junyu Huang, Sarah Klein
My role: UX Designer, Front-End Developer.
You can learn machine learning, too.
People with no prior knowledge of machine learning often perceive that the subject is “pure magic” or a black box. However, it is a systematic way of building programs that learn from data. In this project, we aim to create an interactive tutorial that leverages data visualization to instruct novices about machine learning. Our application will walk users through an image classification example and let them train and test their own image classifiers. In doing so, we will teach users about parts of the machine learning pipeline such as exploratory visualization, model/feature selection, training, testing, and interpreting results.
Contextual Inquiry
Conducting contextual inquiry with a novice in machine learning
Instructional Experience Design
We interviewed both novices and experts in machine learning and learned how they approached an image difference classification problem on deforestation differently. From this, we identified gaps in knowledge that we could focus our instruction on.
Animations for the tutorial section
Besides animations that demonstrate the concepts, we also provide an interactive panel in the tutorial section where students can try building an image recognition classifier with our example data set.
Implementing the Experience
Conducting contextual inquiry with a novice in machine learning
For more, please visit our fully functional tutorial website.