ResNet Sparsifier: Learning Strict Identity Mappings in Deep Residual Networks
Explore this paper's citation graph
Summary
Epsilon-ResNet is proposed that allows us to automatically discard redundant layers, which produces responses that are smaller than a threshold epsilon, with a marginal or no loss in performance.
- Type
- preprint
- Published
- 2018-04-05
- Cited by
- 16
- References
- 56
- Access
- Open access
- OpenAlex
- https://openalex.org/W2795409696
- Semantic Scholar
- https://api.semanticscholar.org/CorpusID:4596535
Keywords
Residual neural network, Residual, Computer science, Artificial intelligence, Deep learning
References
- On the importance of initialization and momentum in deep learning
- Faster R-CNN: Towards Real-Time Object Detection with Region Proposal Networks
- Understanding the difficulty of training deep feedforward neural networks
- Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification
- Some Improvements on Deep Convolutional Neural Network Based Image Classification
- Efficient and accurate approximations of nonlinear convolutional networks
- Sparse Convolutional Neural Networks
- Speeding up Convolutional Neural Networks with Low Rank Expansions
- Going deeper with convolutions
- Algorithms for Hyper-Parameter Optimization
- On the power of small-depth threshold circuits
- Object Detection Networks on Convolutional Feature Maps
- ImageNet Large Scale Visual Recognition Challenge
- Practical Bayesian Optimization of Machine Learning Algorithms
- ImageNet classification with deep convolutional neural networks
- Exploiting Linear Structure Within Convolutional Networks for Efficient Evaluation
- Efficient and Robust Automated Machine Learning
- Deep Residual Learning for Image Recognition
- Quantized Convolutional Neural Networks for Mobile Devices
- Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
Cited by
- Knowledge Representing: Efficient, Sparse Representation of Prior Knowledge for Knowledge Distillation
- Dynamic Residual Dense Network for Image Denoising
- Towards Understanding the Importance of Shortcut Connections in Residual Networks
- A CNN-based methodology for breast cancer diagnosis using thermal images
- A Review on Recent Progress in Thermal Imaging and Deep Learning Approaches for Breast Cancer Detection
- Continuous-in-Depth Neural Networks
- Real-Time Remote Health Monitoring System Driven by 5G MEC-IoT
- Pig Weight and Body Size Estimation Using a Multiple Output Regression Convolutional Neural Network: A Fast and Fully Automatic Method
- Establishment of super sonic inlet flow pattern monitoring system: A workflow
- EfficientWord-Net: An Open Source Hotword Detection Engine based on One-shot Learning
- ISRToken: Learning similarities Tokens for precise infrared spectrum recognition model via Transformer
- Deep Model Compression via Two-Stage Deep Reinforcement Learning
- Intelligent metaheuristic cluster-based wearable devices for healthcare monitoring in telemedicine systems
- Comparison of different approaches to reduce the number of parameters in Deep Neural Networks
- Pixel-Based Attack on ODENet Classifiers
- α-UNet++: A Data-Driven Neural Network Architecture for Medical Image Segmentation
Related papers
- Depth-Specific Variational Scaling Method to Improve Accuracy of ResNet
- IR-UWB radar based through-wall human detection with limited data via multi-hop residual CNN
- Inception-v4, Inception-ResNet and the Impact of Residual Connections on Learning
- Residual network with dense block
- An improved residual network model for image recognition using a combination of snapshot ensembles and the cutout technique
- Multipath Residual Network for Spectral-Spatial Hyperspectral Image Classification