♎
Limited AI
  • Machine Learning
    • Linear Model Cheating Sheet
    • Nonlinear Model Cheating Sheet
    • General Linear Model 1
    • General Linear Model 2
    • General Linear Model 3
    • Tree Based Methods
    • Tree Based Methods Supplement
    • XG,Cat,Light__Boosting
    • KNN&PCA
    • Model Performance
    • Model Evaluation
    • Code Practice
      • KNN
      • Decision Tree Python Code
    • Data and Feature Engineering
      • Handle Bias Data
      • Cold Start Problem
  • Deep Learning
    • Summary v2
    • Basic Neural Network
      • From Linear to Deep
      • Perceptron and Activation Function
      • NN network Details
      • Backpropagation Details
      • Gradient Vanishing vs Gradient Exploding
    • Basic CNN
      • Why CNN
      • Filter/ Convolution Kernel and Its Operation
      • Padding& Stride
      • Layers
      • Extra:From Fully Connected Layers to Convolutions
      • Extra: Multiple Input and Multiple Output Channels
    • Advance CNN
      • Convolutional Neural Networks(LeNet)
      • Deep Convolution Neural Networks(AlexNet)
      • Networks Using Blocks (VGG)
      • Network in Network(NiN)
      • Multi-Branch Networks(GoogLeNet&I mageNet)
      • Residual Networks(ResNet) and ResNeXt
      • Densely Connected Networks(DenseNet)
      • Batch Normalization
    • Basic RNN
      • Seq Model
      • Raw Text to Seq
      • Language Models
      • Recurrent Neural Networks(RNN)
      • Backpropagation Through Time
    • Advance RNN
      • Gated Recurrent Units(GRU)
      • Long Short-Term Memory(LSTM)
      • Bidirectional Recurrent Neural Networks(BRNN)
      • Encoder-Decoder Architecture
      • Seuqence to Sequence Learning(Seq2Seq)
    • Attention Mechanisms and Transformers
      • Queries, Keys, and Values
      • Attention is all you need
        • Attention and Kernel
        • Attention Scoring Functions
        • The Bahdanau Attention Mechanism
        • Multi-Head Attention
        • Self-Attention
        • Attention的实现
      • The Transformer Architecture
        • Extra Reading
        • 最短的最大路径长度
      • Large-Scaling Pretraning with Transformers
        • BERT vs OpenAI GPT vs ELMo
        • Decoder Model框架
        • Bert vs XLNet
        • T5& GPT& Bert比较
        • 编码器-解码器架构 vs GPT 模型
        • Encoder vs Decoder Reference
      • Transformers for Vision
      • Transformer for Multiomodal
    • NLP Pretraining
      • Word Embedding(word2vec)
        • Extra Reading
      • Approximate Training
      • Word Embedding with Global Vectors(GloVe)
        • Extra Reading
        • Supplement
      • Encoder(BERT)
        • BERT
        • Extra Reading
      • Decoder(GPT&XLNet&Lamma)
        • GPT
        • XLNet
          • XLNet架构
          • XLNet特点与其他比较
      • Encoder-Decoder(BART& T5)
        • BART
        • T5
  • GenAI
    • Introduction
      • GenAI Paper Must Read
      • GenAI六个阶段
    • Language Models Pre-training
      • Encoder-Decoder Architecture
      • Encoder Deep Dive
      • Decoder Deep Dive
      • Encoder VS Decoder
      • Attention Mechanism
      • Transformers
    • Example: Llama 3 8B架构
    • Fine-Tuning Generation Models
    • RAG and Adavance RAG
    • AI Agent
  • Statistics and Optimization
    • A/B testing
    • Sampling/ABtesting/GradientMethod
    • Gradient Decent Deep Dive
  • Machine Learning System Design
    • Extra Reading
    • Introduction
  • Responsible AI
    • AI Risk and Uncertainty
      • What is AI risk
      • General Intro for Uncertainty Quantification
      • Calibration
      • Conformal Prediction
        • Review the linear regression
        • Exchangeability
        • Split Conformal Prediction
        • Conformalized Quantile Regression
        • Beyond marginal coverage
        • Split Conformal Classification
        • Full Conformal Coverage
        • Cross-Validation +
        • Conformal Histgram Regression
    • xAI
      • SHAP value
  • Extra Research
    • Paper Reading
    • Reference
Powered by GitBook
On this page
  • 为什么选择卷积神经网络（CNN）？ | Why Choose Convolutional Neural Networks (CNN)?
  • Convolution and Filter in Image Processing
  1. Deep Learning
  2. Basic CNN

Why CNN

为什么选择卷积神经网络（CNN）？ | Why Choose Convolutional Neural Networks (CNN)?

1. 捕捉空间特征 | Captures Spatial Features

CNN 擅长处理图像数据，因为它可以捕捉到数据中的空间结构。通过卷积操作，CNN 可以识别图像中的边缘、纹理和形状等局部特征。 CNN excels at handling image data because it captures spatial structures in the data. Through convolution, CNN can identify local features such as edges, textures, and shapes.

2. 参数更少 | Fewer Parameters

相比全连接网络，CNN 使用共享权重的卷积核，大大减少了需要学习的参数。这让网络更加高效，降低了过拟合的风险。 Compared to fully connected networks, CNN uses shared weights in convolutional filters, significantly reducing the number of parameters to learn. This makes the network more efficient and reduces the risk of overfitting.

3. 平移不变性 | Translation Invariance

卷积操作使得 CNN 对图像中的平移变得不敏感。无论一个物体出现在图像的哪个位置，CNN 都可以识别出来。 The convolution operation allows CNNs to be invariant to translations in the image. CNN can recognize objects regardless of where they appear in the image.

4. 自动特征提取 | Automatic Feature Extraction

CNN 的多层结构可以自动从低级特征（如边缘）到高级特征（如物体）进行提取，无需人工设计特征。 The multi-layer structure of CNNs allows them to automatically extract features from low-level (e.g., edges) to high-level (e.g., objects), eliminating the need for manual feature design.

5. 池化层降低维度 | Dimensionality Reduction via Pooling

池化操作（如最大池化）可以有效地减少数据维度，同时保留关键特征，降低计算成本，并增加对噪声的鲁棒性。 Pooling operations (e.g., max pooling) efficiently reduce data dimensions while preserving key features, lowering computational cost, and making the network more robust to noise.

6. 处理高维数据 | Handles High-Dimensional Data

对于图像这样的高维数据（例如 224x224 的 RGB 图像有超过 150,000 个输入值），CNN 能通过局部连接和池化操作更好地处理这些复杂的输入。 For high-dimensional data like images (e.g., a 224x224 RGB image has over 150,000 input values), CNNs handle this complexity more effectively through local connections and pooling.

Convolution and Filter in Image Processing

  • 图像处理中的卷积操作就是利用卷积核（卷积模板/滤波矩阵/kernel/filter）在图片上滑动，将图像点上的像素灰度值与对应的卷积核上的数值相乘，然后将所有相乘后的值相加作为卷积核中间像素对应的图像上像素的灰度值，并最终滑动完成所有图像的过程。

  • 对图像和卷积核进行逐个元素相乘再求和的操作就相当于将一个二维的函数移动到另一个二维函数的所有位置，这个操作就叫卷积（convolution）或者互相关（cross-correlation）。卷积和相关的差别是，在数学上，卷积需要先对卷积核进行水平和垂直方向的翻转，然后再对前面的矩阵像素相乘求和的操作。但如果矩阵是对称的，那么两者就没有什么差别了。深度学习中的卷积操作在数学上准确度来说称为互相关。

PreviousBasic CNNNextFilter/ Convolution Kernel and Its Operation

Last updated 7 months ago