Flexible, High Performance Convolutional Neural Networks for Image Classification

By Dan Claudiu Ciresan, Ueli Meier, Jonathan Masci, Luca Maria Gambardella, J. Urgen Schmidhuber. 2011

Subfields: Computer Vision and Pattern Recognition, Efficiency & Compression

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Computer Vision and Pattern Recognition

In Efficiency & Compression

Citations in the corpus

Training Very Deep Networks (2015, background): “For instance, within just a few years, the top-5 image classiﬁcation accuracy on the 1000-class ImageNet dataset has increased from ∼ 84% [1] to ∼ 95% [2, 3] using deeper networks with rather small receptive ﬁelds [4, 5].”
Highway Networks (2015, background): “…accuracy on the 1000-class ImageNet dataset has increased from ∼ 84% (Krizhevsky et al., 2012) to ∼ 95% (Szegedy et al., 2014; Simonyan & Zisserman, 2014) through the use of ensembles of deeper architectures and smaller receptive ﬁelds (Ciresan et al., 2011a;b; 2012) in just a few years.”
Very Deep Convolutional Networks for Large-Scale Image Recognition (2014, background): “Second, we decrease the number of parameters: assuming that both the input a d the output of a three-layer 3× 3 convolution stack hasC channels, the stack is parametrised by3 ( 32C2 ) = 27C2 weights; at the same time, a single7 × 7 conv. layer would require72C2 = 49C2 parameters, i.e.81% more.”
End-to-End Training of Deep Visuomotor Policies (2015, background): “, 1989; Schmidhuber, 2015), and have recently gained prominence due to excellent results on a number of vision benchmarks (Ciresan et al., 2011; Krizhevsky et al., 2012; Ciresan et al., 2012; Girshick et al., 2014a; Tompson et al., 2014; LeCun et al., 2015; He et al., 2015).”
Deep Speech: Scaling up end-to-end speech recognition (2014, background): “Researchers have also begun choosing designs that map well to GPU hardware to gain even more efﬁciency, including convolutional [23, 4, 35] and locally connected [7, 5] networks, especially when optimized libraries like cuDNN [2] and BLAS are available.”
Learning Factored Representations in a Deep Mixture of Experts (2013, background): “Deep networks have achieved very good performance in a variety of tasks, e.g. [10, 5, 3].”
Fast image scanning with deep max-pooling convolutional neural networks (2013, methodology): “Pl is obtained as a function of Pl−1 [6].”
Wireless Network Intelligence at the Edge (2018, methodology): “Next, at the pooling layer, the previous layer’s outputs are combined into a single perceptron in the next layer, by selecting their maximum value or taking their average value [76].”
Recent advances in convolutional neural networks (2015, background): “Low-rank matrix factorization has been exploited in a variety of contexts to improve the optimization problems.”
On the Binding Problem in Artificial Neural Networks (2020, background): “Modern neural networks have proven highly successful and superior to symbolic approaches in perceptual domains, such as in visual object recognition (Cireşan et al., 2011, 2012; Krizhevsky et al., 2012) or 1.”
Deep Networks with Internal Selective Attention through Feedback Connections (2014, background): “Deep convolutional neural networks (CNNs) [1] with max-pooling layers [2] trained by backprop [3] on GPUs [4] have become the state-of-the-art in object recognition [5, 6, 7, 8], segmentation/detection [9, 10], and scene parsing [11, 12] (for an extensive review see [13]).”
A comprehensive, application-oriented study of catastrophic forgetting in DNNs (2019, methodology): “CONV A convolutional neural network (CNN) based on the work of Cirean et al. (2011). It is optimized to perform well on image classification problems like MNIST.”
Understanding Deep Architectures using a Recursive Convolutional Network (2013, background): “Convolutional networks have recently made significant progress in a variety of image classification and detection tasks [13, 2, 23], with further gains and applications continuing to be realized.”
On Learning to Think: Algorithmic Information Theory for Novel Combinations of Reinforcement Learning Controllers and Recurrent Neural World Models (2015, methodology): “Today’s faster computers, such as GPUs, mitigate the Fundamental Deep Learning Problem for FNNs [181, 34, 198, 38, 40].”