Unsupervised Learning

Question 1

unsupervised learning

Answer 1

learning algorithm works out the patterns data contains for the purpose of compact representation (dimensionality reduction) and categorisation & analysis (clustering)

Question 2

dimensionality reduction

Answer 2

transform data into a latent space representation of that data

Question 3

latent space

Answer 3

reduced dimensional space in which essential features of input data are encoded

Question 4

latent space in autoencoder

Answer 4

compressed representation of input data

Question 5

encoding

Answer 5

process of transforming/compressing data into latent representation

Question 6

decoding

Answer 6

process of recovering/reconstructing data from latent space compressed

Question 7

auto encoder

Answer 7

MLP (fully connected network) whose first hidden layer constitutes an encoder with the output of layer being the latent representation then the next layer after latent space is decoder

Question 8

how does dimensionality reduction happen with autoencoders

Answer 8

encoding input data into a lower-dimensional space by setting #neurons to be smaller than # of neurons

Question 9

how is network trained with autoencoder

Answer 9

network is trained end to end to minimise MSE (difference between input and reconstructed output)

Question 10

usages of autoencoders

Answer 10

compressing file into a zip
noise reduction

Question 11

auto encoder architecture picture

Answer 11

input and output number same
hourglas shape
more hidden layers allows you to find more features
reduced neuron count in encoder to find compressed latent data (important patterns) then expand neuron count to extract

Question 12

PCA stands for

Answer 12

principal component analysis

Question 13

what is PCA and what is it used for

Answer 13

used for dimensionality reduction while preserving as much variance as possible in the data

It transforms the original dataset into a new coordinate system, where the axes (called principal components) are ordered by the amount of info it conveys about data

Question 14

how to do PCA to compute components

Answer 14

given Nx D matrix
compute mean value for each attribute
compute covariance matrix C
get sorted eigen values
first K rows of E gives K most important components of X

Question 15

applications of PCA

Answer 15

data visualisation (visualise high-dimensional data in 2D or 3D)
noise reduction (retains only the most significant principal components)
feature reduction

Question 16

AE vs PCA

Answer 16

AE
- can handle non-linear data
- more flexible but longer to train
PCA
- only handle linear projections

Question 17

latent variable

Answer 17

variable that is not directly observed but is inferred from other variables that are observed
like intelligence which is inferred from say responses on tests

Question 18

K means algorithm

Answer 18

unsupervised learning algorithm designed to partition data into K distinct clusters
minimises within cluster variance aiming to group similar data points

Question 19

4 step of K-Means algorithm

Answer 19

1) initialise number of clusters K and randomly select K initial clusters
2) assignment
each data point assigned to nearest cluster group by calculating euclidean distance
3) algorithm updates cluster centres as means of each cluster group
4) repeat 2 and 3 until assignment to clusters doesn’t change

Question 20

K means alg is prone

Answer 20

prone to finding local minima

Question 21

how do you find correct # of clusters

Answer 21

run and use heuristic to judge quality of clustering for different choices of K

Question 22

have __ sets of data to mitigate under/over fitting data

Answer 22

training data to train the model
validation data to tune model parameters
test data to test the performance of the model

Question 23

instead of a single validation set you could have

Answer 23

repeated cross validation
The dataset is divided into
k subsets (or folds).
The model is trained on
k-1 and validated on the last one
generates approx set of how well classifier will do on unseen data

Question 24

evaluation metrics for both classification and regression

Answer 24

classification: accuracy, ROC curve (illustrates the trade-off between the true positive rate (sensitivity) and the false positive rate )
regression: MSE and MAE