Microsoft has developed a data visualization tool called TensorWatch to help researchers visualize and debug machine learning models. The increasing complexity and size of machine learning models and datasets can make it difficult for researchers to understand the performance of their model at various stages of training. TensorWatch creates real-time visualizations, including histograms, pie charts, scatter charts, and bar charts. Researchers can also view their data in 3D and compare the results of multiple training runs.
Visualizing Machine Learning Models
Michael McLaughlin is a research assistant at the Center for Data Innovation. He researches and writes about a variety of issues related to information technology and Internet policy, including digital platforms, e-government, and artificial intelligence. Michael graduated from Wake Forest University, where he majored in Communication with Minors in Politics and International Affairs and Journalism. He received his Master’s in Communication at Stanford University, specializing in Data Journalism.
View all posts by Michael McLaughlin