Super Sparse Convolutional Neural Networks
Authors: Yao Lu, Guangming Lu, Bob Zhang, Yuanrong Xu, Jinxing Li4440-4447
AAAI 2019 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | Comparative experiments were performed on the less abundant CIFAR and low resolution Image Net datasets. The results showed that the SSC-Nets can significantly decrease the parameters and the computational Flops without any performance losses. Additionally, it can also improve the ability of addressing the over-fitting problem on the more challenging less abundant datasets. |
| Researcher Affiliation | Academia | 1Department of Computer Science and Technology, Harbin Institute of Technology (Shenzhen), Shenzhen, China 2Department of Computer and Information Science,University of Macau, Macau 3Department of Computing, Hong Kong Polytechnic University, Hung Hom, Kowloon, Hong Kong, China |
| Pseudocode | No | The paper includes equations and architectural descriptions in tables but does not contain explicitly structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not include an explicit statement or link providing access to open-source code for the described methodology. |
| Open Datasets | Yes | CIFAR datasets (Krizhevsky and Hinton 2009)... 50,000 images for training and 10,000 images for testing... Low resolution Image Net datasets (Chrabaszcz, Loshchilov, and Hutter 2017)... |
| Dataset Splits | Yes | CIFAR datasets (Krizhevsky and Hinton 2009) have a small number of images including CIFAR-10 and CIFAR100. They both have 60,000 colored nature scene images in total and the images size is 32 32. There are 50,000 images for training and 10,000 images for testing in 10 and 100 classes. |
| Hardware Specification | No | The paper does not provide specific details about the hardware (e.g., GPU models, CPU specifications) used for running the experiments. |
| Software Dependencies | No | The paper mentions using SGD and Nesterov momentum but does not specify software names with version numbers (e.g., Python, PyTorch, TensorFlow versions). |
| Experiment Setup | Yes | The mini-batch size is set to 128. The SGD method and Nesterov momentum (Sutskever et al. 2013) are utilized in the optimization. Where the momentum is 0.9. On CIFAR, the training epoch is set to 160, and the optimization starts from the initial learning rate with 0.1, which is divided by 10 at the 80th and 120th epoch. For Image Net, the training epoch is 40. The learning rate starts from 0.01 and is divided by 10 every 10 epoches. Finally, the weight decay is set to 0.0002 and 0.0001 on CIFAR and Image Net, respectively. |