Key terms Flashcards
(30 cards)
What is supervised learning?
What is unsupervised learning?
What is unsupervised learning?
What is an example?
A feature-label pair (sometimes, when the context is clear, we may use the term examples to refer to a collection of inputs, even when the corresponding labels are unknown)
types of supervised learning tasks (5)
- Regression
- Classification
- Searching
- Recommending
- Sequence learning
What is the difference between regression and classification?
Regression predicts a continuous value whilst classification predicts a categorical outcome
What is optimisation?
the process of fitting the model to our observed data by altering the model parameters
What is the value produced by the loss function?
How is a particular model parameter updated after calculating the loss?
The original value subtract the derivative of the loss with respect to the model parameter
What is the gradient vector?
A vector that contains all the partial derivates
In order to calculate the gradient with respect to some input features, what must the output be?
A scalar
Common regression problem examples (3)
- Predicting prices
- Predicting length of stay
- Forecasting demand
What is the bias in linear regression?
a value that represents the value of y when all inputs are 0
What do the weights represent in linear regression?
What is gradient descent?
an optimisation algorithm that iteratively reduces the error by updating the model parameters in the direction that incrementally lowers the loss function
Consequence of a learning rate which is too small?
the optimisation process may converge very slowly
Consequence of a learning rate which is too large?
the optimisation process may become unstable
Output dimensions of Xw? (linear regression)
n x 1 (one output for every observation/example)
What is the purpose of the softmax operation?
convert them to probabilities
What is the purpose of activation functions?
to introduce non-linearity into the model, enabling it to learn and represent complex relationships in the data
What exactly do activation functions do?
Decide whether whether a neuron should be activated or not by calculating the weighted sum and further adding bias to it.
Commonly used activation functions (3)
- ReLU
- Sigmoid
- Tanh
What happens if a neuron is not activated?
the neuron does not send its output signal to neurons in the next layer of the neural network
What is generalisation?
the ability of a trained model to perform well on new, unseen data that it hasn’t encountered during training
