Andrew Trask Deep Learning Flashcards
What is the bread and butter fo applied AI/ narrow AI?
Trask 2019
Supervised machine learning
What is the essence of supervised learning?
Trask 19
Taking what you know as input
And quickly transforming it into what you want to know
The majority of work using machine learning results int eh training of what? ]
(Trask 19)
Of a supervised classifier of some kind
What does unsupervised learning share with supervised learning? And what is different?
(Trask 19)
It transforms one dataset into another
But the dataset it transforms into is not previously known or understood
What are some key differences between of unsupervised learning from supervised learning?
(Trask 19)
There is no ‘right answer’ for unsupervised learning
Rather just just tell an unsupervised algorithms to find patters in this data and tell me about them
What is a example of unsupervised learning? And what is the sequence?
(Trask 19)
Clustering a dataset into groups
Clustering transforms a sequence of data points into a sequence of cluster labels
What does unsupervised learning assign the data into labels represented as numbers?
(Trask 19)
Algorithms does know, but rather says hey here is some structure, it looks like there are groups in your data
What is one simple idea of understanding unsupervised learning?
(Trask 19)
Clustering into labels
All forms of unsupervised learning can be viewed as a form of clustering
What does parametricism entail?
Trask 19
Way the learning is stored and often, by extension, the method of learning
Differences between parametric and non parametric models? And example for square in peg
Parametric mdoel
- fixed number of parameters
- trial and error with peg
Nonparametric
- number of parameters is infinite
- tends to count sides to determine where peg goes
