Product development /

Explore our new Knowledge center

February 26 2019/5 min read
  • Ele-Kaja Gildemann
    Ele-Kaja GildemannProduct Owner

The Peltarion Platform keeps improving, functionality by functionality. Let's see what happened to the platform in February!


Support for multi-class classification models

Support for multi-class classification models with higher dimensionality targets is now on the platform. Previously, each row in the dataset had to correspond to exactly one class. We now allow targets of higher dimensionality (e.g., a target that is a vector of different classes, or a target that is an image with one class per pixel). This unlocks use cases such as multi-class semantic segmentation of images.

To train a multi-class target model, the target data can be represented by a numpy array, where the last axis is interpreted as the class label and needs to be one-hot-encoded before importing into the platform.

Visualizations for higher dimensionality targets

Visualizations for higher dimensionality targets on the Evaluation page are now available. Previously, the metrics under Model evaluation were only computed when the target corresponded to exactly one class or to exactly one numeric value. We now also provide the graphs for multi-dimensional targets (e.g., a vector of numeric values or a target image with one class per pixel).

In the case of a classification problem with a multi-dimensional target, the confusion matrix is sampled to a maximum of 500,000 values. For a multi-dimensional target, each value in the confusion matrix corresponds to a vector element, or to a pixel in the target image. This means that the total number of values in the confusion matrix will be many more than the number of samples in the dataset.

In the case of a regression problem with a multi-dimensional target, each dot in the scatter plot represents an element in the target vector or a pixel value in the target image. For visibility reasons, the scatter plot is sampled to show a maximum of 500 data points. The error distribution plot is based on 5,000 sampled values.

Our new Knowledge center – home of everything you might need when exploring the Peltarion Platform

Our new Knowledge center is live!

Home of everything you need when exploring the Peltarion Platform, making sure you get the most out of your experience and the best chance to fulfill your needs for your deep learning projects. Click around, do one of our hands-on platform tutorials, read about and understand different deep learning concepts, explore and learn. Head over to the Knowledge center and start your learning journey!

Haven’t tried out the platform yet? Get started for free now.

  • Ele-Kaja Gildemann

    Ele-Kaja Gildemann

    Product Owner

    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.

Product development topics

02/ More on Product development