6: Tutorial stuff Flashcards
What does nCr expand to?
nCr = n! / r! * (n - r)!
What is a full joint distribution?
???
How do you find missing probabilities in full joint distributions?
???
What is a Bayesian network?
A dag map of probabilistic connections between probabilities such that an arrow points from a probability that is conditionally dependant on another to the othe
How do you build a Bayesian network?
???
How do you draw a Bayesian network?
???
What is a knowledge base (in logic)?
???
What are entailments?
???
When can you infer entailments?
???
What is a most general unifier?
???
What is a unifier?
???
What is unification?
???
What is substitution?
???
How does unification work?
???
What is the factoring rule?
???
How does the factoring rule work?
???
What is an 8-puzzle?
???
What is NY taxi distance?
min non-diagonal node by node distance between two points on a graph (vertical and horizontal moves only)
What is Manhattan distance?
min non-diagonal node by node distance between two points on a graph (vertical and horizontal moves only)
What is backprop?
???
How does backprop work?
???
How does backprop relate to steepest gradient descent?
???
How do you find the error-weight gradient for a unit behind a current hidden one, given its output (input into the current one)?
???
How do gradients come to explode or vanish when the chain rule is applied to multiple hidden layers?
???
What is the connection between validation and early stopping?
early stopping done to mainimise error in training/validation data to improve validation process
What are hyperplanes?
???
Where do hyperplanes rotate to from their initial setting when training is applied through back-propagation?
???
What are convolutional neural networks?
Divide and conquer ish neural networks that convolute data to find patterns before recombining adjacent data into groups. often used in computer vision.
How have convolutional neural networks helped recent breakthroughs in machine learning?
allowed advances in computer vision
What is the difference in capability between a recurrent neural network and a feedforward network? Where does this difference arise from?
recurrent facilitates context and memory because of neurons being able to connect to themselves
