Current browse context:
stat.ML
Change to browse by:
References & Citations
Statistics > Machine Learning
Title: Estimate of the Neural Network Dimension using Algebraic Topology and Lie Theory
(Submitted on 6 Apr 2020 (v1), last revised 8 Mar 2022 (this version, v12))
Abstract: In this paper we present an approach to determine the smallest possible number of neurons in a layer of a neural network in such a way that the topology of the input space can be learned sufficiently well. We introduce a general procedure based on persistent homology to investigate topological invariants of the manifold on which we suspect the data set. We specify the required dimensions precisely, assuming that there is a smooth manifold on or near which the data are located. Furthermore, we require that this space is connected and has a commutative group structure in the mathematical sense. These assumptions allow us to derive a decomposition of the underlying space whose topology is well known. We use the representatives of the $k$-dimensional homology groups from the persistence landscape to determine an integer dimension for this decomposition. This number is the dimension of the embedding that is capable of capturing the topology of the data manifold. We derive the theory and validate it experimentally on toy data sets.
Submission history
From: Luciano Melodia [view email][v1] Mon, 6 Apr 2020 14:15:05 GMT (462kb,D)
[v2] Fri, 19 Jun 2020 09:32:13 GMT (501kb,D)
[v3] Tue, 14 Jul 2020 13:09:50 GMT (503kb,D)
[v4] Sat, 1 Aug 2020 08:58:54 GMT (503kb,D)
[v5] Thu, 22 Oct 2020 06:17:04 GMT (503kb,D)
[v6] Wed, 28 Oct 2020 08:45:41 GMT (503kb,D)
[v7] Tue, 3 Nov 2020 15:02:55 GMT (503kb,D)
[v8] Tue, 10 Nov 2020 14:29:01 GMT (505kb,D)
[v9] Mon, 16 Nov 2020 09:14:42 GMT (505kb,D)
[v10] Sun, 13 Dec 2020 14:36:48 GMT (505kb,D)
[v11] Thu, 31 Dec 2020 07:27:13 GMT (505kb,D)
[v12] Tue, 8 Mar 2022 20:59:48 GMT (505kb,D)
Link back to: arXiv, form interface, contact.