ResNet: Enabling Deep Convolutional Neural Networks through Residual Learning

📅 2025-10-27
📈 Citations: 0
Influential: 0
📄 PDF

career value

166K/year
🤖 AI Summary
To address the training difficulty of ultra-deep convolutional neural networks (CNNs) caused by vanishing gradients, this paper proposes a residual learning framework. It introduces identity-mapping shortcut connections that allow input signals to bypass certain nonlinear transformations, thereby ensuring unimpeded forward signal propagation and backward gradient flow. This approach enables stable end-to-end training of CNNs with over 100 layers—e.g., ResNet-110—on CIFAR-10, achieving 89.9% accuracy, significantly surpassing comparably deep conventional CNNs (84.1%) while exhibiting faster convergence and greater training robustness. The core innovation lies in reformulating the learning objective from approximating a target mapping to learning a residual function, synergistically combined with batch normalization. This fundamentally alleviates optimization challenges inherent in very deep networks and establishes a foundational paradigm for deep model architecture design.

Technology Category

Application Category

📝 Abstract
Convolutional Neural Networks (CNNs) has revolutionized computer vision, but training very deep networks has been challenging due to the vanishing gradient problem. This paper explores Residual Networks (ResNet), introduced by He et al. (2015), which overcomes this limitation by using skip connections. ResNet enables the training of networks with hundreds of layers by allowing gradients to flow directly through shortcut connections that bypass intermediate layers. In our implementation on the CIFAR-10 dataset, ResNet-18 achieves 89.9% accuracy compared to 84.1% for a traditional deep CNN of similar depth, while also converging faster and training more stably.
Problem

Research questions and friction points this paper is trying to address.

Overcoming vanishing gradient problem in deep networks
Enabling training of very deep convolutional neural networks
Using skip connections for direct gradient flow
Innovation

Methods, ideas, or system contributions that make the work stand out.

Skip connections enable deep network training
Residual learning overcomes vanishing gradient problem
Shortcut connections allow direct gradient flow
🔎 Similar Papers
No similar papers found.