andy
acb7dc429e
Affected files: .obsidian/graph.json .obsidian/workspace-mobile.json .obsidian/workspace.json STEM/AI/Neural Networks/Architectures.md STEM/AI/Neural Networks/CNN/CNN.md STEM/AI/Neural Networks/CNN/Examples.md STEM/AI/Neural Networks/CNN/FCN/FCN.md STEM/AI/Neural Networks/CNN/GAN/DC-GAN.md STEM/AI/Neural Networks/CNN/GAN/GAN.md STEM/AI/Neural Networks/CNN/Interpretation.md STEM/AI/Neural Networks/Deep Learning.md STEM/AI/Neural Networks/MLP/MLP.md STEM/AI/Neural Networks/SLP/Least Mean Square.md STEM/AI/Neural Networks/Transformers/Attention.md STEM/AI/Neural Networks/Transformers/Transformers.md STEM/img/feedforward.png STEM/img/multilayerfeedforward.png STEM/img/recurrent.png STEM/img/recurrentwithhn.png
1.3 KiB
1.3 KiB
Before 2010s
- Data hungry
- Need lots of training data
- Processing power
- Niche
- No-one cared/knew about CNNs
After
- ImageNet
- 16m images, 1000 classes
- GPUs
- General processing GPUs
- CUDA
- NIPS/ECCV 2012
- Double digit % gain on ImageNet accuracy
Full Connected
- Move from convolutional operations towards vector output
- Stochastic drop-out
- Sub-sample channels and only connect some to MLP layers
As a Descriptor
- Most powerful as a deeply learned feature extractor
- MLP classifier at the end isn't fantastic
- Use SVM to classify prior to penultimate layer
Finetuning
- Observations
- Most CNNs have similar weights in conv1
- Most useful CNNs have several conv layers
- 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
- Learning rate = 0
- Randomly initialise remaining layers
- Continue with existing weights
Training
- Validation & training Deep Learning#Loss Function
- Early
- Under-fitting
- Training not representative
- Later
- Overfitting
- V.Deep Learning#Loss Function can help adjust learning rate
- Or indicate when to stop training