stem/AI/Neural Networks/CNN/CNN.md
andy 5f167f25a4 vault backup: 2023-06-04 22:30:39
Affected files:
.obsidian/app.json
.obsidian/workspace-mobile.json
.obsidian/workspace.json
STEM/AI/Neural Networks/CNN/CNN.md
STEM/AI/Neural Networks/CNN/GAN/cGAN.md
STEM/AI/Neural Networks/MLP/Back-Propagation.md
2023-06-04 22:30:39 +01:00

54 lines
1.6 KiB
Markdown

## Before 2010s
- Data hungry
- Need lots of training data
- Processing power
- Niche
- No-one cared/knew about CNNs
## After
- [ImageNet](../CV/Datasets.md#ImageNet)
- 16m images, 1000 classes
- GPUs
- General processing GPUs
- CUDA
- NIPS/ECCV 2012
- Double digit % gain on [ImageNet](../CV/Datasets.md#ImageNet) accuracy
# Full Connected
[Dense](../MLP/MLP.md)
- Move from [convolutional](Convolutional%20Layer.md) operations towards vector output
- Stochastic drop-out
- Sub-sample channels and only connect some to [dense](../MLP/MLP.md) layers
# As a Descriptor
- Most powerful as a deeply learned feature extractor
- [Dense](../MLP/MLP.md) classifier at the end isn't fantastic
- Use SVM to classify prior to penultimate layer
![cnn-descriptor](../../../img/cnn-descriptor.png)
# Finetuning
- Observations
- Most CNNs have similar weights in [conv1](Convolutional%20Layer.md)
- Most useful CNNs have several [conv layers](Convolutional%20Layer.md)
- Many weights
- Lots of training data
- Training data is hard to get
- Labelling
- Reuse weights from other network
- Freeze weights in first 3-5 [conv layers](Convolutional%20Layer.md)
- Learning rate = 0
- Randomly initialise remaining layers
- Continue with existing weights
![fine-tuning-freezing](../../../img/fine-tuning-freezing.png)
# Training
- Validation & training [loss](../Deep%20Learning.md#Loss%20Function)
- Early
- Under-fitting
- Training not representative
- Later
- Overfitting
- V.[loss](../Deep%20Learning.md#Loss%20Function) can help adjust learning rate
- Or indicate when to stop training
![under-over-fitting](../../../img/under-over-fitting.png)