Dataset view: Redesign and feature preparation improvements — Product update
This month, platform updates are focused on usability and flexibility, relating to the dataset preparation moment of your AI creation journey. These are the latest changes made to the Peltarion Platform:
Flexibility
The Dataset view has been redesigned! Feature settings are now available on the Inspector panel to the right of your workspace and are shown upon selecting a specific feature. This way, all configurations are always available for you in the same location, making it easier to find and adjust what’s needed. This change also allows for more working space in the dataset area, showing more data examples.
Furthermore, in the feature settings, you can edit the proposed feature label and set the appropriate way of encoding the feature before feeding it into the neural network, or leave it without making any adjustments. The platform provides suitable encoding options based on the data types of your features.
Missing the one-hot encoded option under Dataset features? This is now called Categorical and can be used for both text and numbers. When choosing Categorical as encoding, the feature will be automatically translated into numeric features across all the classes that are found in the dataset, this is needed to transform categories into suitable format for neural networks. For numeric features (including images and float tensors), standardization and min-max encoding options are available, to make sure numeric values are scaled into meaningful ranges that improve your neural network training and make it more accurate and faster.
Usability
Combined feature capability added to the Dataset view. This means users no longer need to form feature sets in order to use them as input and target features for a model. The Combined feature option, however, still enables combining many features into a single feature and makes it easier to feed this combined feature as input to the model. All the feature shape calculations are done automatically by the platform.
Those were some of the updates made this month. Stay tuned for more exciting news in May!
Check out our latest webinars or sign up for our upcoming ones here if you want to know more about how the Peltarion Platform can be used in practice and go through real-life use cases within different application areas.
Originally published at peltarion.com.
Author
Ele-Kaja Gildemann is a Product Owner at Peltarion. She has a degree in computer science from Tallinn University of Technology and more than 15 years of experience in sectors as diverse as digital services, telecom and retail. She is passionate about data-driven product development, user experience and machine learning.
