My machine learning research is focused on methods that utilize froms of unsupervised learning. In this context I am fascinated by prototype-based, topology representing networks that can learn the topological structure of an underlying input space [1], i.e., they identify dense manifolds of data in otherwise sparse, high-dimensional space. A method that I am particularly interested in is the growing neural gas (GNG) algorithm that was introduce by Bernd Fritzke [2]. Over the years I developed multiple variants of this algorithm, e.g., the recursive growing neural gas (RGNG) [3], the differential growing neural gas (DGNG) [4], or the local input space histograms (LISH) extension [5].
References
1
,
Topology representing networks,
In: Neural Networks, 7:507–522, 1994,
[doi]
2
,
A growing neural gas network learns topologies,
In: Advances in Neural Information Processing Systems 7, pp. 625–632, MIT Press, 1994,
[details]
3
,
A Computational Model of Grid Cells based on a Recursive Growing Neural Gas,
In: PhD thesis. University of Hagen, 2016,
[pdf|bibtex]
4
,
Efficient Approximation of a Recursive Growing Neural Gas,
In: Computational Intelligence: International Joint Conference, IJCCI 2018 Seville, Spain, September 18-20, 2018 Revised Selected Papers, 2021,
[pdf|doi|bibtex]
5
,
Supporting GNG-based clustering with local input space histograms,
In: Proceedings of the 22nd European Symposium on Artificial Neural Networks, Computational Intelligence and Machine Learning. Louvain-la-Neuve, Belgique, pp. 559–564, 2014,
[pdf|bibtex]