Deep Learning in Finance Lect7 Flashcards
Deep Learning led to the concept of?
Artificial Neural Networks - using layers of artificial neurons to receive input and apply an activation function along with a human set threshold.
In single layer neural network, there consists what factors?
- model inputs - multiple real numbers,
- Weights, which connect inputs to unit
- Single model unit
- Output
Model inputs in Single Layer neural networks are usually associated with
real numbers
Explain the importance and detail of weights
Weights need high computational power to optimize “w” values, connect input to model to output
Output is in Single Layer Neural Networks consist of?
Weighted sum of inputs plus a bias term b, only achieved when input exceeds a threshold
3 common activation functions
Identity: f(a) = a
(upwards straight line)
Sigmoid: f(a) = 1/ 1+e^-a (curve)
ReLu: f(a) = max(0,a)
Hidden layer is the act of?
Misdirecting weights to another layer
Neural network
Parametreized family of non-linear functions
Types of Neural network
- feed forward (acyclic) - single-layer or multi-layer
- recurrent (cyclic)
Key Idea of Neural network Learning
Adjusting the weights, changes the function represented by the neural network
(learning = optimisation in weight space)
Difference between target output and network output
Iteratively adjust weight to reduce error
Anything in between input and output layer is called
Hidden layer