[1512.00567] Rethinking the Inception Architecture for Computer Visionopen searchopen navigation menucontact arXivsubscribe to arXiv mailings

Convolutional networks are at the core of most state-of-the-art computer vision solutions for a wide variety of tasks. Since 2014 very deep convolutional networks started to become mainstream, yielding substantial gains in various benchmarks. Although increased model size and computational cost tend to translate to immediate quality gains for most tasks (as long as enough labeled data is provided for training), computational efficiency and low parameter count are still enabling factors for various use cases such as mobile vision and big-data scenarios. Here we explore ways to scale up networks in ways that aim at utilizing the added computation as efficiently as possible by suitably factorized convolutions and aggressive regularization. We benchmark our methods on the ILSVRC 2012 classification challenge validation set demonstrate substantial gains over the state of the art: 21.2% top-1 and 5.6% top-5 error for single frame evaluation using a network with a computational cost of 5 bill

1 mentions: @imenurok
Date: 2020/07/04 05:21

Referring Tweets

@imenurok Label Smoothingは後の研究で結構出てくるけど、元論文だとそこまで大きな扱いじゃなかったりする t.co/ecYXSKjEjK

Bookmark Comments

Related Entries

Read more GitHub - yasunorikudo/chainer-DenseNet: Densely Connected Convolutional Network implementation by Ch...
2 users, 0 mentions 2018/12/26 02:14
Read more [1805.00932] Exploring the Limits of Weakly Supervised Pretraining
2 users, 3 mentions 2019/06/24 03:48
Read more GitHub - pytorch/ignite: High-level library to help with training neural networks in PyTorch
2 users, 1 mentions 2019/08/06 05:17
Read more GitHub - maciejkula/rustlearn: Machine learning crate for Rust
2 users, 1 mentions 2020/08/12 13:02
Read more GitHub - microsoft/onnxjs: ONNX.js: run ONNX models using JavaScript
4 users, 0 mentions 2020/09/18 14:22