Dynamic Clustering Convolutional Neural Network
Authors: Tanzhe Li, Baochang Zhang, Jiayi Lyu, Xiawu Zheng, Guodong Guo, Taisong Jin
AAAI 2025 | Venue PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | The extensive experiments of image classification, object detection, instance segmentation, and semantic segmentation on the benchmark datasets demonstrate that the proposed DCCNe Xt outperforms the mainstream Convolutional Neural Networks (CNNs), Vision Transformers (Vi Ts), Vision Multi-layer Perceptrons (MLPs), Vision Graph Neural Networks (GNNs), and Vision Mambas (Vi Ms). |
| Researcher Affiliation | Academia | 1Key Laboratory of Multimedia Trusted Perception and Effcient Computing, Ministry of Education of China, Xiamen University, China. 2School of Informatics, Xiamen University, China. 3Hangzhou Research Institute, School of Artificial Intelligence, Beihang University, China. 4 Nanchang Institute of Technology, China. 5 School of Engineering Science, University of Chinese Academy of Sciences, China. 6 Ningbo Institute of Digital Twin, Eastern Institute of Technology, Ningbo, China. 7 Key Laboratory of Oracle Bone Inscriptions Information Processing, Ministry of Education of China, Anyang Normal University, China. EMAIL,EMAIL, EMAIL |
| Pseudocode | No | The paper describes the method using figures, equations, and prose, but does not include a clearly labeled pseudocode or algorithm block. |
| Open Source Code | Yes | Code https://github.com/ltzovo/DCCNe Xt |
| Open Datasets | Yes | We use Image Net-1K (Russakovsky et al. 2015) to conduct the image classification experiments. Image Net-1K (Russakovsky et al. 2015) is also called the ISLVRC 2012 dataset, with 1K classes containing 1.28M training images and 50K validation images. |
| Dataset Splits | Yes | Image Net-1K (Russakovsky et al. 2015) is also called the ISLVRC 2012 dataset, with 1K classes containing 1.28M training images and 50K validation images. |
| Hardware Specification | No | The paper does not provide specific details about the hardware used, such as GPU or CPU models. It only mentions the use of PyTorch and Timm for implementation. |
| Software Dependencies | No | The paper mentions using Py Torch (Paszke et al. 2019), Timm (Wightman et al. 2019), MMDetection (Chen et al. 2019), and MMSEG (Contributors 2020), but does not provide specific version numbers for these software dependencies. |
| Experiment Setup | Yes | For a fair comparison, we follow the experimental parameter settings that are widely used in Dei T (Touvron et al. 2021a) and train the proposed model on a 224x224-resolution image with 300 epochs. We employ the Adam W (Loshchilov and Hutter 2017) optimizer using a cosine decay learning rate scheduler with 20 epochs of linear warm-up. Data augmentation and regularization techniques include Rand Augmentation (Cubuk et al. 2020), Mixup (Zhang et al. 2017), Cut Mix (Yun et al. 2019), Random Erasing (Zhong et al. 2020), Weight Decay, Label Smoothing (Szegedy et al. 2016), and Stochastic Depth (Huang et al. 2016). |