The cascading neural network: building the Internet of Smart Things

By Sam Leroux, Steven Bohez, Elias De Coninck, Tim Verbelen, Bert Vankeirsbilck, Pieter Simoens, Bart Dhoedt. 2017

58 citations recorded by the source corpus.

Subfields: Cognitive Neuroscience, Computer Vision and Pattern Recognition, Electrical and Electronic Engineering, Model Architectures, ML Systems & Hardware

How researchers see this paper

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

In Cognitive Neuroscience

1. Jiwen Lu
2. Simone Scardapane
3. Yongming Rao
4. Jie Zhou

In Computer Vision and Pattern Recognition

1. Jie Zhou
2. Jiwen Lu
3. Shiji Song
4. Gao Huang
5. Luc Van Gool
6. Yulin Wang
7. Dacheng Tao
8. Yi Yang

In Electrical and Electronic Engineering

1. Shiji Song
2. Yulin Wang
3. Liekang Zeng
4. Tailin Liang
5. Gao Huang
6. C. John Glossner
7. Zhi Zhou
8. Diana Marculescu

In Model Architectures

1. Shiji Song
2. Yulin Wang
3. Liekang Zeng
4. Tailin Liang
5. Gao Huang
6. Ting-Wu Chin
7. C. John Glossner
8. Zhi Zhou

In ML Systems & Hardware

1. Xu Chen
2. Diana Marculescu
3. Dimitrios Stamoulis
4. Liekang Zeng
5. Gao Huang
6. Zhi Zhou
7. Ofer Dekel
8. Ji Lin

Citations in the corpus

• Dynamic Neural Networks: A Survey (2021, background): “based criteria [45], [51] or learned functions [46], [52], [53].”
• Why Should We Add Early Exits to Neural Networks? (2020, methodology): “In this case, an estimation of the cost of communicating across the channel can be included both in the design of the early exits, as described in “Placement and Optimized Design of the Early Exits,” or in the model selection/regularization phase, as described in “Regularizing for Computational Cost”
• Adaptive Neural Networks for Efficient Inference (2017, positive): “Others have developed cascade approaches (Leroux et al., 2017; Odena et al., 2017) that allow early exits based on confi-dence feedback.”
• Designing Adaptive Neural Networks for Energy-Constrained Image Classification (2018, background): “network [15], and the cascading neural network [16].”
• Runtime Neural Pruning (2017, methodology): “Cascade methods [26, 28, 39, 40] relied on manually-selected thresholds to control execution.”
• Pruning and Quantization for Deep Neural Network Acceleration: A Survey (2021, methodology): “The cascading neural network by Leroux [136] reduced the average inference time of overfeat network [201] by 40% within 2% ILSVRC-2012 top-1 accuracy drop.”
• Edge Intelligence: Paving the Last Mile of Artificial Intelligence With Edge Computing (2019, background): “For example, the cascading network [91] simply adds the MP layer and fully connected layer to the standard DNN model and achieves a speedup of 20%.”