A Neural Network That Learns to Do Hyphenation

Fritzke, Bernd; Nasahl, Christof

doi:10.1016/b978-0-444-89178-5.50092-0

Cited by 74 publications

(115 citation statements)

References 1 publication

Supporting

Mentioning

106

Contrasting

Unclassified

Order By: Relevance

“…The Growing Neural Gas (GNG) algorithm presented in (Fritzke, 1995) is a vector-based learning method which is capable of building (clustering) topological relationships that are present in the input space by applying the Competitive Hebbian Learning (CHL) rule. The neurons in the m-dimensional continuous output space R m are competing for the right to respond to a given input signal ξ with ξ ∈ R n of the discrete input training dataset D. A single input signal contains three coordinates of a point in the input space.…”

Section: Building a Topological Model Via The Growing Neural Gas Algomentioning

confidence: 99%

“…This method (Equation 3) saturates N after k steps to avoid too many performance dips and to force the network to a consistent distribution of the nodes in R m . The other hyperparameters which are not listed in the table are set to the default values from (Fritzke, 1995). Nevertheless, a small increase of tmax is necessary to keep the network error low with the increasing amount of data.…”

Section: Hyperparameter Adaptionmentioning

confidence: 99%

See 1 more Smart Citation

Next-Best-View Method Based on Consecutive Evaluation of Topological Relations

Dierenbach¹,

Weinmann²,

Jutzi³

2016

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

ABSTRACT:This work describes an iterative algorithm for estimating optimal viewpoints, so called next-best-views (NBVs). The goal is to incrementally construct a topological network from the scene during the consecutive acquisition of several views. Our approach is a hybrid method between a surface-based and a volumetric approach with a continuous model space. Hence, a new scan taken from an optimal position should either cover as much as possible from the unknown object surface in one single scan, or densify the existing data and close possible gaps. Based on the point density, we recover the essential and structural information of a scene based on the Growing Neural Gas (GNG) algorithm. From the created graph representation of topological relations, the density of the point cloud at each network node is estimated by approximating the volume of Voronoi cells. The NBV Finder selects a network node as NBV, which has the lowest point density. Our NBV method is self-terminating when all regions reach a predefined minimum point density or the change of the GNG error is zero. For evaluation, we use a Buddha statue with a rather simple surface geometry but still some concave parts and the Stanford Dragon with a more complex object surface containing occluded and concave parts. We demonstrate that our NBV method outperforms a "naive random" approach relying on uniformly distributed sensor positions in terms of efficiency, i.e. our proposed method reaches a desired minimum point density up to 20% faster with less scans.

show abstract

Section: Building a Topological Model Via The Growing Neural Gas Algomentioning

confidence: 99%

Section: Hyperparameter Adaptionmentioning

confidence: 99%

Next-Best-View Method Based on Consecutive Evaluation of Topological Relations

Dierenbach¹,

Weinmann²,

Jutzi³

2016

Int. Arch. Photogramm. Remote Sens. Spatial Inf. Sci.

View full text Add to dashboard Cite

show abstract

“…With Growing Neural Gas (GNG) [10] method a growth process takes place from minimal network size and new neurons are inserted successively using a particular type of vector quantization. To determine where to insert new neurons, local measures are gathered during the adaptation process and each new neuron is inserted close to the neuron with highest accumulated error.…”

Section: Gng Algorithmmentioning

confidence: 99%

“…However, as the NG has a fixed number of nodes, it is necessary to have some a priori information about the input space to pre-establish the size of the network. This model was extended by [10] proposing the Growing Neural Gas (GNG) network, which combined the flexible structure of the NG with a growing strategy. Moreover, the learning adaptation step was slightly modified.…”

Section: Introductionmentioning

confidence: 99%

3D Surface Reconstruction of Noisy Point Clouds Using Growing Neural Gas: 3D Object/Scene Reconstruction

Orts-Escolano

García-Rodríguez

Morell

et al. 2015

Neural Process Lett

View full text Add to dashboard Cite

With the advent of low-cost 3D sensors and 3D printers, scene and object 3D surface reconstruction has become an important research topic in the last years. In this work, we propose an automatic (unsupervised) method for 3D surface reconstruction from raw unorganized point clouds acquired using low-cost 3D sensors. We have modified the Growing Neural Gas (GNG) network, which is a suitable model because of its flexibility, rapid adaptation and excellent quality of representation, to perform 3D surface reconstruction of different real-world objects and scenes. Some improvements have been made on the original algorithm considering colour and surface normal information of input data during the learning stage and creating complete triangular meshes instead of basic wire-frame representations. The proposed method is able to successfully create 3D faces online, whereas existing 3D reconstruction methods based on Self-Organizing Maps (SOMs) required postprocessing steps to close gaps and holes produced during the 3D reconstruction process. A set of quantitative and qualitative experiments were carried out to validate the proposed method. The method has been implemented and tested on real data, and has been found to be effective at reconstructing noisy point clouds obtained using low-cost 3D sensors.

show abstract

“…We therefore use a growing-when-required (GWR) network [15] to efficiently cluster keypoints that are highly similar. A GWR-network is a clustering method, very similar to a growing-neural-gas (GNG) network [7]. Both networks are based on Kohonen's self-organizing maps (SOM) [10].…”

Section: Keypoint Clusteringmentioning

confidence: 99%

Active exploration and keypoint clustering for object recognition

Kootstra

Ypma

Boer

2008

2008 IEEE International Conference on Robotics and Automation

View full text Add to dashboard Cite

Abstract-Object recognition is a challenging problem for artificial systems. This is especially true for objects that are placed in cluttered and uncontrolled environments. To challenge this problem, we discuss an active approach to object recognition. Instead of passively observing objects, we use a robot to actively explore the objects. This enables the system to learn objects from different viewpoints and to actively select viewpoints for optimal recognition. Active vision furthermore simplifies the segmentation of the object from its background. As the basis for object recognition we use the Scale Invariant Feature Transform (SIFT). SIFT has been a successful method for image representation. However, a known drawback of SIFT is that the computational complexity of the algorithm increases with the number of keypoints. We discuss a growing-whenrequired (GWR) network for efficient clustering of the keypoints. The results show successful learning of 3D objects in real-world environments. The active approach is successful in separating the object from its cluttered background, and the active selection of viewpoint further increases the performance. Moreover, the GWR-network strongly reduces the number of keypoints.

show abstract

A Neural Network That Learns to Do Hyphenation

Cited by 74 publications

References 1 publication

Next-Best-View Method Based on Consecutive Evaluation of Topological Relations

Next-Best-View Method Based on Consecutive Evaluation of Topological Relations

3D Surface Reconstruction of Noisy Point Clouds Using Growing Neural Gas: 3D Object/Scene Reconstruction

Active exploration and keypoint clustering for object recognition

Contact Info

Product

Resources

About