Daniël Haasbroek
Also known as Daniël G. Haasbroek, Daniël Gerbrand Haasbroek, Gerbrand Haasbroek
Daniël Haasbroek is an AI researcher who studies how neural networks learn and generalize. He coauthored a 2020 study of binary node activations and published the work in SACAIR proceedings in 2021. In 2024, he coauthored a study of network fragmentation, shared first as an arXiv preprint and included in the SciForDL workshop at NeurIPS. As of October 2026, he is a PhD student at North-West University with an MEng in Computer and Electronic Engineering, researching generalisation in deep convolutional networks through information flow.
Timeline
-
2020
Binary node activations
In 2020, he and Marelie H. Davel study how binary node activations group training samples during neural-network learning. Their experiments find that activations in deeper layers become more sensitive to class membership as training progresses.
-
2021
A SACAIR conference paper
In 2021, he publishes the study with Marelie H. Davel in the Southern African Conference for Artificial Intelligence Research proceedings.
-
November 2024
A fragmentation preprint
In November 2024, he and five coauthors submit “Is network fragmentation a useful complexity measure?” to arXiv. The study examines network fragmentation as a complexity measure for image classification and tests its link to generalization performance.
-
2024
A NeurIPS workshop paper
The study appears in the Scientific Methods for Understanding Deep Learning workshop at NeurIPS in 2024. It reports that fragmentation-based measures perform well on the PGDL generalization benchmark.
-
As of October 2026
PhD student at North-West University
As of October 2026, he is a PhD student at North-West University with an MEng in Computer and Electronic Engineering. His research theme is generalisation in deep convolutional neural networks using information flow.
Sources and credits
Photo: We have no published photo of Daniël Haasbroek yet. Send us a free-licensed photo.