1,744 Hits in 5.6 sec

Parsimonious random vector functional link network for data streams

Mahardhika Pratama, Plamen P. Angelov, Edwin Lughofer, Meng Joo Er
2018 Information Sciences  
inconsequential hidden nodes can be pruned and input features can be dynamically selected. pRVFLN puts into perspective an online active learning mechanism which expedites the training process and relieves  ...  A novel class of RVLFN, namely parsimonious random vector functional link network (pRVFLN), is proposed in this paper. pRVFLN features an open structure paradigm where its network structure can be built  ...  guarantees a compact and parsimonious network structure and the rule recall mechanism which aims to overcome the cyclic concept drift. pRVFLN incorporates an online feature selection scenario which is  ... 
doi:10.1016/j.ins.2017.11.050 fatcat:na2bkrp6gzamfgunqihekqeghu

Optimization and deployment of CNNs at the edge

Paolo Meloni, Luca Benini, Maura Pintor, Battista Biggio, Bernhard Moser, Natalia Shepeleva, Nikos Fragoulis, Ilias Theodorakopoulos, Michael Masin, Francesca Palumbo, Daniela Loi, Paola Busia (+6 others)
2019 Proceedings of the 16th ACM International Conference on Computing Frontiers - CF '19  
The proposed tool flow aims at automating different design steps and reducing development costs.  ...  This requires deployment of DL algorithms on low-energy and resource-constrained computing nodes, often heterogenous and parallel, that are usually more complex to program and to manage without adequate  ...  ACKNOWLEDGMENTS This work has received funding from the European Union's Horizon 2020 Research and Innovation Programme under grant agreement No. 780788.  ... 
doi:10.1145/3310273.3323435 dblp:conf/cf/MeloniLBDPSSM0B19 fatcat:axfii7pnincxvkvxyxtub6nhpm


P. Meloni, O. Ripolles, D. Solans, M. Pintor, B. Biggio, T. Stefanov, S. Minakova, N. Fragoulis, I. Theodorakopoulos, M. Masin, F. Palumbo, D. Loi (+7 others)
2018 Proceedings of the Workshop on INTelligent Embedded Systems Architectures and Applications - INTESA '18  
This requires implementation on low-energy computing nodes, often heterogenous and parallel, that are usually more complex to program and to manage.  ...  The framework introduces architecture-awareness, considering the target inference platform very early, already during algorithm selection, and driving the optimal porting of the resulting embedded application  ...  The authors would like to thank all the participants taking part in the project for their support, including Giuseppe Desoli and Giulio Urlini from STMicroelectronics srl, Adriano Souza Ribeiro and Werner  ... 
doi:10.1145/3285017.3285019 dblp:conf/esweek/MeloniLDPSMSCBR18 fatcat:b2rri4rd3nfuvety3rsx73kviu

Model structure determination in neural network models

Humberto M. Henrique, Enrique L. Lima, Dale E. Seborg
2000 Chemical Engineering Science  
This procedure is based on network pruning using the orthogonal least-squares technique to determine insigni"cant or redundant synaptic weights, biases, hidden nodes and network inputs.  ...  The results also show the importance of pruning procedures to identify parsimonious FNN models.  ...  Classical pruning techniques Parsimonious structures for FNNs can be obtained by the process of pruning, that consists in deleting unnecessary weights or/and nodes, starting with a large network and going  ... 
doi:10.1016/s0009-2509(00)00170-6 fatcat:dzwlaex6szd7vjsv2im5bkmtdi

Balancing Robustness against the Dangers of Multiple Attractors in a Hopfield-Type Model of Biological Attractors

Ron C. Anafi, Jason H. T. Bates, Jean Peccoud
2010 PLoS ONE  
By progressively removing the links of fully connected Hopfield nets, we found that a designated attractor of the nets could still be supported until only slightly more than 1 link per node remained.  ...  Conclusions/Significance: We speculate that homeostatic biological networks may have evolved to assume a degree of connectivity that balances robustness and agility against the dangers of becoming trapped  ...  To characterize this connectivity more fully, we pruned 10 different 200-node networks and found that the final network configurations had between 201 and 207 connections in total.  ... 
doi:10.1371/journal.pone.0014413 pmid:21203505 pmcid:PMC3008716 fatcat:faw3dvoxnbhyffup23bajn3jfa

Improving the human readability of features constructed by genetic programming

Matthew Smith, Larry Bull
2007 Proceedings of the 9th annual conference on Genetic and evolutionary computation - GECCO '07  
Genetic Programming is combined with a Genetic Algorithm to construct and select new features from those available in the data, a potentially significant process for data mining since it gives consideration  ...  We then examine techniques to improve the human readability of these new features and extract more information about the domain.  ...  They used the GA to perform feature selection for a face recognition dataset where feature subsets were evaluated through their use by C4.5.  ... 
doi:10.1145/1276958.1277291 dblp:conf/gecco/SmithB07 fatcat:apxelrqnfjbr5cxajwtcddfvdy

Generalization Techniques for Layered Neural Networks in the Classification of Remotely Sensed Images

1999 Doboku Gakkai Ronbunshu  
With this discussion, we provide a feasible technique to design the LNN in consideration of its generalization capability.  ...  Finally, we apply the proposed technique to a practical land cover classification using remotely sensed images, and demonstrate its potential.  ...  Concerning the architecture design, the size of network (the number of layers and nodes) and the type of activation functions are important factors.  ... 
doi:10.2208/jscej.1999.618_95 fatcat:altwkajqi5eltihife6ycs47ii

Designing Energy-Efficient Arithmetic Operators Using Inexact Computing

Avinash Lingamneni, Christian Enz, Krishna Palem, Christian Piguet
2013 Journal of Low Power Electronics  
This design philosophy of inexact computing is of particular interest in the domain of embedded and (portable) multimedia applications and in application domains of budding interest such as recognition  ...  sustained technology scaling prophesied by the Moore's Law would end within the next decade or so, attributed primarily to an understanding that switching devices can no longer function deterministically as feature  ...  The authors would also express their sincere thanks to Jean-Luc Nagel, Pierre-Alain Beuchat and Marc Morgan of CSEM SA for their help with the chip tapeout and measurements.  ... 
doi:10.1166/jolpe.2013.1249 fatcat:mpt3p6wz6vdm7lcl7ir3haimru

An application of pruning in the design of neural networks for real time flood forecasting

Giorgio Corani, Giorgio Guariso
2005 Neural computing & applications (Print)  
one at a time, designing a much more parameter-parsimonious model.  ...  We propose the application of pruning in the design of neural networks for hydrological prediction.  ...  Molari and N. Quaranta, Civil Protection Service of Regione Lombardia, for supplying the data of Olona river, M. Campolo and A. Soldati, University of Udine, for the data of Tagliamento river.  ... 
doi:10.1007/s00521-004-0450-z fatcat:emxsoizdszaifgvfzmzyvl2ai4

Efficient in-situ image and video compression through probabilistic image representation [article]

Rongjie Liu, Meng Li, Li Ma
2020 arXiv   pre-print
approaches---including JPEG, JPEG2000, BPG, MPEG4, HEVC and a neural network-based method---for all of these different image types and on nearly all of the individual images and videos over some methods  ...  CARP uses an optimal permutation of the image pixels inferred from a Bayesian probabilistic model on recursive partitions of the image to reduce its effective dimensionality, achieving a parsimonious representation  ...  Recall that all descendants of a pruned node are pruned by design; hence, for each j,k that is pruned, we randomly select a direction d 0 ∈ D( j,k ) and set j+1,2k = A (d 0 ) l and j+1,2k+1 = A (d  ... 
arXiv:1912.05622v3 fatcat:vrdo6v64vzdurluivy73dtld3i

Deep Stacked Stochastic Configuration Networks for Lifelong Learning of Non-Stationary Data Streams [article]

Mahardhika Pratama, Dianhui Wang
2019 arXiv   pre-print
This paper proposes deep stacked stochastic configuration network (DSSCN) for continual learning of non-stationary data streams which contributes two major aspects: 1) DSSCN features a self-constructing  ...  methodology of deep stacked network structure where hidden unit and hidden layer are extracted automatically from continuously generated data streams; 2) the concept of SCN is developed to randomly assign  ...  Acknowledgement This work is fully supported by Ministry of Education, Republic of Singapore, Tier 1 Research Grant and NTU Start-up Grant.  ... 
arXiv:1808.02234v3 fatcat:djemtonbb5ghnorhoijcke447y

Designing High Strength Multi-phase Steel for Improved Strength–Ductility Balance Using Neural Networks and Multi-objective Genetic Algorithms

Shubhabrata Datta, Frank Pettersson, Subhas Ganguly, Henrik Saxén, Niruopam Chakraborti
2007 ISIJ International  
KEY WORDS: high strength multiphase steel; neural network model; pruning algorithm; genetic algorithm; multi-objective optimization; predator prey algorithm; alloy design. Fig. 2.  ...  Two different methods of reducing the network connectivity, viz a pruning algorithm and a multi-objective predator prey genetic algorithm, have been used for neural network modeling of the mechanical properties  ...  In our case, it is difficult to choose a single network from so many reasonably good candidates developed through the pruning or the predator prey algorithms.  ... 
doi:10.2355/isijinternational.47.1195 fatcat:ffteqowjfnetzldxs7tqo7yeny

Using Radial Basis Function Networks for Function Approximation and Classification

Yue Wu, Hui Wang, Biaobiao Zhang, K.-L. Du
2012 ISRN Applied Mathematics  
We also compare the features and capability of the two models.  ...  In this paper, we give a comprehensive survey on the RBF network and its learning.  ...  Acknowledgments The authors acknowledge Professor Chi Sing Leung Department of Electronic Engineering, City University of Hong Kong and Professor M. N. S.  ... 
doi:10.5402/2012/324194 fatcat:bwdbufult5dqdjpotnezuu7hei

Logic Explained Networks [article]

Gabriele Ciravegna, Pietro Barbiero, Francesco Giannini, Marco Gori, Pietro Lió, Marco Maggini, Stefano Melacci
2021 arXiv   pre-print
In this paper, we propose a general approach to Explainable Artificial Intelligence in the case of neural architectures, showing how a mindful design of the networks leads to a family of interpretable  ...  The large and still increasing popularity of deep learning clashes with a major limit of neural network architectures, that consists in their lack of capability in providing human-understandable motivations  ...  Acknowledgments We thank Ben Day, Dobrik Georgiev, Dmitry Kazhdan, and Alberto Tonda for useful feedback and suggestions.  ... 
arXiv:2108.05149v1 fatcat:iovnkfoofzg35c7qv4y3wtp2na

Criminal Community Detection Based on Isomorphic Subgraph Analytics

Theyvaa Sangkaran, Azween Abdullah, NZ Jhanjhi
2020 Open Computer Science  
Community structure, generally described as densely connected nodes and similar patterns of links is an important property of complex networks.  ...  We studied community detection in criminal networks using the graph theory and formally introduced an algorithm that opens a new perspective of community detection compared to the traditional methods used  ...  to gain a general overview of the criminal network graph, and second to apply the relevant methods and techniques designed such as pruning, and newly developed community detection algorithm with isomorphic  ... 
doi:10.1515/comp-2020-0112 fatcat:ayulyinzyfhlnaxlyevrep7fn4
« Previous Showing results 1 — 15 out of 1,744 results