Login Register



Auto-login on future visits

Forgot your password?

perClass 4 announcement

perClass 4 grow bigger

We are happy to announce perClass 4 major release!


perClass grew to be simpler and smarter. The GUI tools got more
user-friendly. Classifiers are now easier to use and more intelligent.
That means you get better results with less typing.

See the summary of improvements in the release notes.

New: Work with databases!

Embedded SQLite engine

Do you want to classify data stored in a database? Good news: perClass 4 comes with a new DB option. It enables you to:

  • Work with nominal features and train classifiers on categorical and mixed representations. More info
  • Leverage multi-gigabyte local databases. Use SQL queries to create data sets directly from Matlab. More info
  • Deploy high-throughput classifiers to MS SQL server via stored procedures with perClass Connector.
Bring decision-making to your database server!

perClass Toolbox becomes modular.

NEW Imaging and DB options

  • Base: General machine learning tools, feature extraction, interactive visualization and performance optimization tools.
  • DB Option: Work with nominal features and SQL databases. Learn more
  • Imaging Option: Extract local image features. Work with image regions of arbitrary shape. Learn more.

perClass course: Machine Learning for R&D Specialists.

Announcement

training course on industrial machine learning

Want to learn how to build the best classifier for your problem?
Join us on 28 Oct - 1 Nov 2013 in Delft, The Netherlands.

This intensive training course provides you with the practical methodology to develop your own solutions.
"Thanks again for the excellent course. I gained the right knoweldge to delve futher into our machine learning problems." -- Stefan, R&D Leica
More testimonials

100 EUR discount for early bird registration before 31.8.2013.

Learn more register now

How to identify the typical and rare examples of a class?

Tutorial

how to identify typical samples using cascaded rejection

When executing a classifier, each sample gets assigned into one class. But what if we want to know more? For instance, is this observation a typical or a rare example of its class?
In this tutorial, you will learn how to compose multiple classifiers to answer such question.

Watch this 4 minute video to find out more.

Neural Network now scalable to large data sets.

Feature highlight

An advantage of the Neural Network (NN) classifier is that it does not assume or impose any data distribution. Instead, it learns to discriminate classes from the data directly adapting to the problem.
See the scalability of our new NN implementation However, in practice, NN is often discarded due to long training times. Not anymore! perClass 4 brings a new fast NN implementation that scales to millions of samples.
As a result, we now often find it performing among the best classifiers in our industrial projects.
Try it on yours! Read more