PC-DARTS: Partial Channel Connections for Memory-Efficient Differentiable Architecture Search

By Yuhui Xu, Lingxi Xie, Xiaopeng Zhang, Xin Chen, Guo-Jun Qi, Qi Tian, H. Xiong. 2019

150 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.

1. Xu Tan
2. Tao Qin
3. Tie-Yan Liu
4. Zitong Yu
5. Renqian Luo
6. Guohao Li
7. Bernard Ghanem
8. Matthias Müller

Citations in the corpus

• Semi-Supervised Neural Architecture Search (2020, background): “[18, 17, 15, 5, 36, 4, 27, 9] also leverage the one-shot idea to perform efficient search while using different search algorithms.”
• DARTS+: Improved Differentiable Architecture Search with Early Stopping (2019, methodology): “We remark that recent progresses of DARTS, including P-DARTS [4], Auto-DeepLab [19] and PC-DARTS [37], also adopt the early stopping idea implicitly where fewer search epochs are manually set in their methods.”
• SGAS: Sequential Greedy Architecture Search (2019, methodology): “In Differentiable Architecture Search (DARTS) [32] and its variants [52, 8, 54], the optimal architecture is derived from a discrete search space by relaxing the selection of operations to a continuous optimization problem.”
• Neural Architecture Search with GBDT (2020, methodology): “We limit the FLOPS of the discovered architecture to be less than 600M for fair comparison with other works [38, 18, 11, 25, 30, 3] and train it for 300 epochs following [2].”
• Differentiable Feature Aggregation Search for Knowledge Distillation (2020, methodology): “Among the one-shot methods, the differentiable architecture search (DARTS) [5,22,24,26,36,40] further relaxes the discrete search space to be continuous and couples the architecture parameters with the model parameters in the supernet.”
• Adversarially Robust Neural Architectures (2020, unknown): “Boosting NAS algorithms involve optimization on architecture to obtain better performance with small model size [18, 19].”
• Retinex-inspired Unrolling with Cooperative Prior Architecture Search for Low-light Image Enhancement (2020, methodology): “The above search algorithms have achieved highly competitive performance in various high-level vision tasks, including classification [39], detection [33], and segmentation.”
• Searching Central Difference Convolutional Networks for Face Anti-Spoofing (2020, methodology): “Different from previous works [35, 60], we adopt only one output of the latest incoming cell as the input of the current cell.”