2024. 1. 20. 08:11ㆍ코딩 도구/LG Aimers
LG Aimers: AI전문가과정 4차
Module 6. 『딥러닝(Deep Learning)』
ㅇ 교수 : KAIST 주재걸 교수
ㅇ 학습목표
Neural Networks의 한 종류인 딥러닝(Deep Learning)에 대한 기본 개념과 대표적인 모형들의 학습원리를 배우게 됩니다.
이미지와 언어모델 학습을 위한 딥러닝 모델과 학습원리를 배우게 됩니다.
Part 3. Convolutional Neural Networks and Image Classification
-Various Neural Network Architectures
Fully Connected Neural Network
ConvNets or CNN (Convolutional Neural Network)
RNN (Recurrent Neural Network)
-Challenges in Computer Vision Tasks
다 같은 고양이라는 걸 학습시켜야한다.
-Filtering: the Math Behind the Match
1. Overlap the convolution filter and the image patch.
2. Multiply each image pixel by the corresponding filter coefficient.
3. Add them up.
4. Divide by the total number of pixels in the feature. (optional)
-Convolution Layer
• A stack of images becomes filtered images by a bunch of stacked filters
-Pooling: Shrinking the Image Stack
1. Pick a window size (usually 2).
2. Pick a stride (usually 2).
3. Walk your window across your filtered images.
4. From each window, take the maximum value.
-Rectified Linear Units (ReLUs)
-Deep Stacking
• Layers can be repeated several (or many) times
-Backpropagation
• Error = right answer – actual answer
-Hyperparameters
Convolution
• Number of filters
• Size of filters
Pooling
• Window size
• Window stride
Fully Connected
• Number of layers
• Number of neurons
-Advanced CNN Architectures
Various CNN Architectures
• AlexNet (시초 2012)
• VGGNet
• GoogLeNet
• ResNet
-Summary: CNN Architectures
• VGG, GoogLeNet, and ResNet are all in wide use
• ResNet is currently the best default
• Recent trends are going towards extremely deep networks
'코딩 도구 > LG Aimers' 카테고리의 다른 글
LG Aimers 4기 Transformer (4) | 2024.01.22 |
---|---|
LG Aimers 4기 Seq2Seq , Natural Language Understanding and Generation (2) | 2024.01.21 |
LG Aimers 4기 Training Neural Networks (3) | 2024.01.19 |
LG Aimers 4기 Deep Neural Networks (2) | 2024.01.18 |
LG Aimers 4기 Phase 1 온라인 교육 후기 (0) | 2024.01.17 |