One of the observations made in earth data science is the massive increase of data volume (e.g, higher resolution measurements) and dimensionality (e.g. hyper-spectral bands). Traditional data mining tools (Matlab, R, etc.) are becoming redundant in the analysis of these datasets, as they are unable to process or even load the data. Parallel and scalable techniques, though, bear the potential to overcome these limitations. In this contribution we therefore evaluate said techniques in a High Performance Computing (HPC) environment on the basis of two earth science case studies: (a) Density-based Spatial Clustering of Applications with Noise (DBSCAN) for automated outlier detection and noise reduction in a 3D point cloud and (b) land cover type classification using multi-class Support Vector Machines (SVMs) in multispectral satellite images. The paper compares implementations of the algorithms in traditional data mining tools with HPC realizations and 'big data' technology stacks. Our analysis reveals that a wide variety of them are not yet suited to deal with the coming challenges of data mining tasks in earth sciences.
In the last century, the costs of powering datacenters have increased so quickly
AbsslrocbThe RealNeumn is a tme of mikine neumn based construction Drocess could be imdemented bv an omimisation ~~ on the descriptions of biological neirolls a i d is-simple enough to be simulated on standard desktop personal computers in a reasonable h e . A mathematid model is presented of the RealNeuron that is suited for optimisation purposes. A RealNeumn eorrerlly, has short term a,,d long term msmn." algorithm that operates on the of the model. 11. MESSAGE PROCESSING The RealNeuron cannot interact directly with the environment and needs a bodv ( E ) to do that. The body limits the if ...-... ".,. environmental states ( % E ) that could be sensed i d affected I. INTRODUCTION by the RealNeuron network. See [I31 for examples from real Artificial neural networks (A"$ have achieved some world applications of such bodies. success in non-linear forecasting, pattern matching and in Only the processing of exchanged messages inside the body artificial life paradigms, [1]-[4]. A, , ANN exhibits the caps. is considered This is not a restriction because the externally bility of redundancy and vague representation of information. exchanged messages U s e a predefined co"udcation Protocol However, an ANN still lacks many of the vital features of to communicate with the environment. The communication a biological neural network (BNN), such as the ability of protocol is implemented through the sensors and actuators biological neurons to allow self-modification with regard to objects that is part of the body definition. short term and long-term learning.In the description of the R" and the R" connections, simulation of B~S developed by neuro-biologists does dendrites are not described. A COmpariSOn with the biological not seem to be promising for engineering applications. Recent neuron may seem as a drawback. The main functions of the attempts have shown that simulations of brain cells wn~umes dendrite in case of the biological neuron are a vast amount of computer resources ( [SI reported that it took (i) the enhancement of the surface of the neuron to be able 18.2h on 5 connected SUN SPARC-2 workstations to simulate one second of activity of a single biological neuron.) (ii) the logical processing of input signals by grouping certain By analyzing the recent neuro-biological literature, [6]-[8], a model of the human brain cell has been derived, called the There exists no spatial restrictions for the R" except for RealNeuron (RN), which encompasses the main features of memory inside a computer. The spacial signal processing can biological neuronal cells that are considered to be the key be realized by defining post-synaptic membrane groups that features of learning, 191. This model is is empirically more are processed together and delayed the relevant amount of sound than the classical A " s and the simulation is running time.in real time or faster on a standard personal computer (PC). Doeben-Henisch has shown that the RealNeuron is a specialA. instance of a spiking neuron [IO]. The message processing (or signal processing)...
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
customersupport@researchsolutions.com
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
This site is protected by reCAPTCHA and the Google Privacy Policy and Terms of Service apply.
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.