MLG 025 Convolutional Neural Networks
Oct 30, 2017
Convnets or CNNs. Filters, feature maps, window/stride/padding, max-pooling.
Show Notes
See resources on Deep Learning episode.
- One-time donations w/ BTC / PayPal
- Image recognition, classification - computer vision
- ML takeover
- Final main network (MLP, RNN, CNN)
- Don't use MLP for images, use CNNs
- Filters -> feature maps -> convolutional layers
- Window, stride, padding
- Max-pooling
- Architectures (ILSVRC ImageNet Challenge)
- LeNet-5
- AlexNet
- GoogLeNet
- Inception
- Resnet
- etc..