GPU implementation of neural networks

Oh, Kyoungsu; Jung, Keechul

doi:10.1016/j.patcog.2004.01.013

Cited by 326 publications

(165 citation statements)

References 2 publications

Supporting

Mentioning

161

Contrasting

Unclassified

Order By: Relevance

“…It is not even necessary that this data be from other robots, as shown by Yang's use of general-purpose cooking videos for object and grasp recognition [79]. Regarding training time, local parallel processing [17] and increases in raw processing speed have led to significant improvements. Distributed computing offers the potential to direct more computing resources to a given problem [88] but can be limited by communication speeds [2].…”

Section: Resultsmentioning

confidence: 99%

See 1 more Smart Citation

Deep learning in robotics: a review of recent research

2017

View full text Add to dashboard Cite

Advances in deep learning over the last decade have led to a flurry of research in the application of deep artificial neural networks to robotic systems, with at least thirty papers published on the subject between 2014 and the present. This review discusses the applications, benefits, and limitations of deep learning vis-à-vis physical robotic systems, using contemporary research as exemplars. It is intended to communicate recent advances to the wider robotics community and inspire additional interest in and application of deep learning in robotics.

show abstract

Section: Resultsmentioning

confidence: 99%

“…In the 2000s, researchers began using graphical processing units (GPUs) to parallelize implementations of artificial neural networks [17]. The largest bottleneck in training neural networks is a matrix-vector multiplication step, which can be parallelized using GPUs.…”

Section: A Brief History Of Deep Learningmentioning

confidence: 99%

Deep learning in robotics: a review of recent research

2017

View full text Add to dashboard Cite

show abstract

“…(Figure 2.) Oh and Jung [6] and Luo et al [7] both treat neural networks as matrix operations and report considerable speed ups by using a GPU when the neural network is used on compute intensive image processing tasks. For example, Ribeiro et al [8] demonstrated speed ups in the region of 170 fold when using Multiple Back-Propagation to predict bankruptcy.…”

Section: Neural Networkmentioning

confidence: 99%

Graphics processing units and genetic programming: an overview

Langdon

2011

Soft Comput

View full text Add to dashboard Cite

A top end graphics card (GPU) plus a suitable SIMD interpreter, can deliver a several hundred fold speed up, yet cost less than the computer holding it. We give highlights of AI and computational intelligence applications in the new field of general purpose computing on graphics hardware (GPGPU). In particular we survey genetic programming (GP) use with GPU. We give several applications from Bioinformatics and show how the fastest GP is based on an interpreter rather than compilation. Finally using GP to generate GPU CUDA kernel C++ code is sketched.

show abstract

“…Among them, the naïve Bayes classifier assumes the data distribution to be Gaussian and finds the discriminant function using prior probability and likelihood. Finally, it derives the decision curve using the discriminant function [2]. However, this method presumes that all of the features are independent, and thus, errors can occur in the process of obtaining the discriminant function [3].…”

Section: Introductionmentioning

confidence: 99%

A Novel Method for a Reliable Classifier using Gradients

Han¹,

Cha²

2017

IEIE Transactions on Smart Processing and Computing

View full text Add to dashboard Cite

Abstract:In this paper, we propose a new classification method to complement a naïve Bayesian classifier. This classifier assumes data distribution to be Gaussian, finds the discriminant function, and derives the decision curve. However, this method does not investigate finding the decision curve in much detail, and there are some minor problems that arise in finding an accurate discriminant function. Our findings also show that this method could produce errors when finding the decision curve. The aim of this study has therefore been to investigate existing problems and suggest a more reliable classification method. To do this, we utilize the gradient to find the decision curve. We then compare/analyze our algorithm with the naïve Bayesian method. Performance evaluation indicates that the average accuracy of our classification method is about 10% higher than naïve Bayes.

show abstract

GPU implementation of neural networks

Cited by 326 publications

References 2 publications

Deep learning in robotics: a review of recent research

Deep learning in robotics: a review of recent research

Graphics processing units and genetic programming: an overview

A Novel Method for a Reliable Classifier using Gradients

Contact Info

Product

Resources

About