Frequency-Domain Dynamic Pruning for Convolutional Neural Networks
By Zhenhua Liu, Jizheng Xu, Xiulian Peng, Ruiqin Xiong. 2018
165 citations recorded by the source corpus.
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
Citations in the corpus
- Similarity-Preserving Knowledge Distillation (2019, methodology): “Weight pruning methods [11, 20, 22, 24, 33, 35, 39] remove unimportant weights from the network, sparsifying the network connectivity structure.”
- Gradient Projection Memory for Continual Learning (2021, neutral): “Let’s consider a Conv layer with the input tensor X ∈ RCi×hi×wi and filters W ∈ RCo×Ci×k×k.”
- FcaNet: Frequency Channel Attention Networks (2020, methodology): “There are also some applications in the model compression and pruning tasks like [6, 27, 38].”
- Filter Pruning via Geometric Median for Deep Convolutional Neural Networks Acceleration (2018, background): “Pruning-based approaches aim to remove the unnecessary connections of the neural network [11, 21, 24].”
- Structured Pruning for Deep Convolutional Neural Networks: A Survey (2023, background): “Pruning Domain: Utilizing representation in the frequency domain [65], [195] to guide pruning is another interesting direction.”
- Federated learning for medical image analysis: A survey (2023, background): “Based on previous studies that low-frequency components of parameters form the basis of deep network capability (Liu et al., 2018), only these low-frequency components are aggregated to share knowledge learned from different clients, while the high-frequency parts are disregarded.”
- Model Compression and Hardware Acceleration for Neural Networks: A Comprehensive Survey (2020, background): “The spatial weights can also be converted to the weight coefficients in the frequency domain and then dynamically pruning the unimportant coefficients can help achieve a higher compression ratio [199].”