Difference between revisions of "Strategy & Tactics"
| Line 4: | Line 4: | ||
*[http://www.deeplearningpatterns.com/doku.php The Deep Learning AI Playbook | Carlos E. Perez] | *[http://www.deeplearningpatterns.com/doku.php The Deep Learning AI Playbook | Carlos E. Perez] | ||
*[http://www.deeplearningbook.org/ Deep Learning | Ian Goodfellow and Yoshua Bengio and Aaron Courville] | *[http://www.deeplearningbook.org/ Deep Learning | Ian Goodfellow and Yoshua Bengio and Aaron Courville] | ||
| + | |||
| + | Architecture: | ||
| + | * Variables type | ||
| + | * Variable scaling | ||
| + | * Cost function | ||
| + | * Type of neural network - CNN, RNN, FFN | ||
| + | * Number of layers, hidden | ||
| + | * Type of layers | ||
| + | ** LSTM, Dense, Highway | ||
| + | ** Convolutional | ||
| + | ** Pooling | ||
| + | * Weight initialization type | ||
| + | * Number of nodes | ||
| + | * Type of activation function - linear, Sigmoid, ReLU | ||
| + | * Dropout rate (or not) | ||
| + | * Threshold | ||
<youtube>dfvycjYIMPo</youtube> | <youtube>dfvycjYIMPo</youtube> | ||
Revision as of 19:50, 13 May 2018
- Neural Network Zoo | Fjodor Van Veen
- The Deep Learning AI Playbook | Carlos E. Perez
- Deep Learning | Ian Goodfellow and Yoshua Bengio and Aaron Courville
Architecture:
- Variables type
- Variable scaling
- Cost function
- Type of neural network - CNN, RNN, FFN
- Number of layers, hidden
- Type of layers
- LSTM, Dense, Highway
- Convolutional
- Pooling
- Weight initialization type
- Number of nodes
- Type of activation function - linear, Sigmoid, ReLU
- Dropout rate (or not)
- Threshold
