AlexNet
4 days ago
- #Image Classification
- #Computer Vision
- #Deep Learning
- AlexNet is a convolutional neural network (CNN) developed for image classification, achieving prominence in the 2012 ImageNet Large Scale Visual Recognition Challenge (ILSVRC).
- Developed by Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton, it was the first widely recognized deep CNN for large-scale visual recognition.
- AlexNet contains eight layers: five convolutional layers and three fully connected layers, utilizing ReLU activation and dropout regularization.
- Trained on 1.2 million images using two Nvidia GTX 580 GPUs, it achieved a top-5 error rate of 15.3%, significantly outperforming competitors.
- AlexNet's success was enabled by large-scale datasets (ImageNet), GPU computing, and improved training methods, marking a turning point in computer vision.
- The architecture influenced subsequent models like GoogLeNet, VGGNet, and ResNet, and its paper has been cited over 172,000 times.
- Key innovations included data augmentation, local response normalization, and parallel GPU training to handle the model's 60 million parameters.