Topological approaches to deep learning

DOI10.1007/978-3-030-43408-3_5MaRDI QIDQ5118362zbMATH OpenOpenAlexFDO

Authors Gunnar Carlsson, Rickard Brüel Gabrielsson

Publication date 8 September 2020

Published in Topological Data Analysis (Search for Journal in Brave)

Full work available at URL https://arxiv.org/abs/1811.01122

Mathematics Subject Classification ID

Persistent homology and applications, topological data analysis (55N31) Artificial neural networks and deep learning (68T07)

Abstract: We perform topological data analysis on the internal states of convolutional deep neural networks to develop an understanding of the computations that they perform. We apply this understanding to modify the computations so as to (a) speed up computations and (b) improve generalization from one data set of digits to another. One byproduct of the analysis is the production of a geometry on new sets of features on data sets of images, and use this observation to develop a methodology for constructing analogues of CNN's for many other geometries, including the graph structures constructed by topological data analysis.

Recommendations

• The Modern Mathematics of Deep Learning
• Learning representations of persistence barcodes
• Deep learning Gauss-Manin connections
• On the expressive power of deep architectures
• An introduction to the mathematics of deep learning

Cites work

• A Klein-bottle-based dictionary for texture representation
• Deep learning
• On the local behavior of spaces of natural images
• The nonlinear statistics of high-contrast patches in natural images
• Topological signal processing
• Topology and data

Cited in

(15)

• Some geometrical and topological properties of DNNs' decision boundaries
• Deep learning Gauss-Manin connections
• A differential geometric approach to time series forecasting
• Deep learning the hyperbolic volume of a knot
• A topological insight into restricted Boltzmann machines
• Shapley homology: topological analysis of sample influence for neural networks
• Topology of deep neural networks
• ToFU: topology functional units for deep learning
• Algebraic dynamical systems in machine learning
• Quantification on the generalization performance of deep neural network with Tychonoff separation axioms
• Functional loops: monitoring functional organization of deep neural networks using algebraic topology
• Trainable and explainable simplicial map neural networks
• A topological data analysis based classifier
• Algebraic machine learning with an application to chemistry
• Topological scale framework for hypergraphs

Describes a project that uses

Uses Software

• MNIST
• CIFAR
• ImageNet

This page was built for publication: Topological approaches to deep learning

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5118362)