Neural Architecture Search with Reinforcement Learning
Citations
22 citations
Cites background from "Neural Architecture Search with Rei..."
...A typical NAS, such as that in [33], is composed of a controller and a trainer....
[...]
...Existing works have demonstrated that the automatically searched neural architectures can achieve close accuracy to the best human-invented architectures [33,34]....
[...]
...Recently, Neural Architecture Search (NAS) has been consistently breaking the accuracy records in a variety of machine learning applications, such as image classification [33], image segmentation [18], video action recognition [22], and many more....
[...]
22 citations
22 citations
Cites methods from "Neural Architecture Search with Rei..."
...Initial NAS methods used evolutionary algorithms and reinforcement learning to find optimal network architectures, but each iteration of the search fully trains and evaluates many child networks [13, 34], thus needing a huge amount of computation resources and time....
[...]
22 citations
Cites background from "Neural Architecture Search with Rei..."
...Tuning the hyperparameters of the first category alone has led to a separate field of research called Neural Architecture Search (NAS) [25] that allowed achievement of state-ofthe-art performance [53, 65] on some benchmark problems, although at a massive computational cost of 800 GPUs for a few weeks....
[...]
...For example, the HPO problem can be seen as reinforcement learning [12, 65, 66] where the main difference between each method relies on how the agents are defined and dealt with....
[...]
22 citations
Cites methods from "Neural Architecture Search with Rei..."
...2 Efficient Neural Architecture Search We search architectures for binary networks by adopting ideas from neural architecture search (NAS) methods for floating point networks [27, 32, 45, 51, 52]....
[...]
References
123,388 citations
111,197 citations
55,235 citations
"Neural Architecture Search with Rei..." refers methods in this paper
...Along with this success is a paradigm shift from feature designing to architecture designing, i.e., from SIFT (Lowe, 1999), and HOG (Dalal & Triggs, 2005), to AlexNet (Krizhevsky et al., 2012), VGGNet (Simonyan & Zisserman, 2014), GoogleNet (Szegedy et al., 2015), and ResNet (He et al., 2016a)....
[...]
42,067 citations
31,952 citations
"Neural Architecture Search with Rei..." refers methods in this paper
...Along with this success is a paradigm shift from feature designing to architecture designing, i.e., from SIFT (Lowe, 1999), and HOG (Dalal & Triggs, 2005), to AlexNet (Krizhevsky et al., 2012), VGGNet (Simonyan & Zisserman, 2014), GoogleNet (Szegedy et al., 2015), and ResNet (He et al., 2016a)....
[...]