Portal/TA3/benchmark suites

Benchmark Suites

Benchmark suites in machine learning are collections of datasets that are commonly used to evaluate the performance of machine learning algorithms. These suites are often used to compare the performance of different algorithms on the same datasets and to identify the strengths and weaknesses of algorithms. Some examples of benchmark suites in machine learning include the MNIST handwritten digit dataset, the CIFAR image classification dataset, and the ImageNet dataset for image classification tasks.

In TA3 - Statistics and Machine Learning - we are developing such benchmark suites for supervised regression, graphical modelling, and neuroimaging and electro- and optophysiology.

Supervised Regression

We have developed a benchmark suite for supervised regression comprising 40 well curated supervised regression problems on tabular datasets The regression suite has been published on OpenML, which is an open platform for sharing machine learning research data.