Deep Learning (CV) - Papers
This category is used to document the reading and reproduction insights of important papers in deep learning for computer vision.
Guanlong Ma
Deep Learning (CV) - Introduction
This category is used to organize deep learning study notes and project practices in the field of computer vision.
Machine Learning - Introduction
This category is used to document fundamental knowledge, methodological practices, and experience summaries related to machine learning.
Deep Learning - Useful Modules
Deep Learning - Activation Functions
Deep Learning - Weight Initialization
1. Overview of Parameter Initialization Neural network models generally rely on the stochastic gradient descent optimizer to update their parameters, and parameter learning is a non-convex problem...
Deep Learning - Loss Functions
1. Overview of Loss Functions Most deep learning algorithms involve some form of optimization. Optimization means changing \(x\) in order to minimize or maximize some function \(f(x)\); we usually...
Deep Learning - Optimization Algorithms
Preface The optimization algorithms of deep neural networks are the algorithms used to update network parameters and minimize the loss function. Optimization algorithms matter a great deal in deep...
Deep Learning - Normalization Layers
1. Internal Covariate Shift Training deep neural networks is complicated because changes in the parameters of earlier layers cause the distribution of each layer’s input to change during training....
Deep Learning - Models
1. MLP (1) Linear layer Definition: A Linear layer is a basic layer type in deep learning frameworks such as PyTorch. It performs a linear transformation, that is, the output is the product of th...