In this video, we're going to talk about more general, non-sequential models in Keras using the functional API.
So there are two types of models in Keras. We've seen the sequential already. There is another type called model, which you would use if you were in need of non-sequential models. Once defined, we will see how to do that in just a second. The model can be trained and evaluated exactly like sequential models. So you don't have to learn anything about that part.
Using the functional API, you would usually start with the model and configure one or several inputs. Once you have those inputs defined, we define transformations for those inputs, until you arrive at one or several outputs.
The initial lectures series on this topic can find in the below links:
Cloud Architectures for AI and Deep Learning
[ Ссылка ]
Linear algebra
[ Ссылка ]
Deep feedforward neural networks
[ Ссылка ]
Convolutional Neural Networks
[ Ссылка ]
Recurrent neural networks
[ Ссылка ]
LSTMs
[ Ссылка ]
Autoencoders and representation learning
[ Ссылка ]
Methods for neural network training
[ Ссылка ]
Gradient Descent Updater Strategies
[ Ссылка ]
How to choose the correct activation function
[ Ссылка ]
The bias-variance tradeoff in deep learning
[ Ссылка ]
Introduction to TensorFlow
[ Ссылка ]
Neural Network Debugging with TensorBoard
[ Ссылка ]
Automatic Differentiation
[ Ссылка ]
Introduction to Keras
[ Ссылка ]
Keras overview
[ Ссылка ]
Sequential models in Keras
[ Ссылка ]
Feed-forward networks
[ Ссылка ]
Recurrent neural networks
[ Ссылка ]
