Technical Interview

Question 1

What is backpropagation?

Answer 1

A technique for training neural networks that uses gradient desent to calculate the loss function at output and distribute it back through the neural network, resulting in adjusted weights for neurons.

Question 2

What is a nonparametric model?

Answer 2

A non-parametric model is one with no assumptions about a variable’s distribution. Easier to use but decreased accuracy.

Question 3

Difference between statistics and machine learning?

Answer 3

Statistics establishes causal relations while machine learning focuses on predictive accuracy.

Question 4

Support Vector Machine is used to create a line between two groups and is more memory intensive and will take longer to tune.

Answer 4

Random forest is for a semi-supervised learning problem such as dissimilarity measure.

Question 5

What is an autoencoder?

Answer 5

type of neural network and a form of feature extraction

Question 6

Bagging and boosting in an ensemble model

Answer 6

Bagging and boosting both will take random samples from the data over and over to help mold better results. In bagging you always have the same probability to appear as data but in boosting it keeps some data to guide the new data set

Question 7

What is a loss function in a neural network?

Answer 7

The loss function describes the magnitude of error a network made during prediction.

Question 8

RDBMS

Answer 8

Relational database management system which maintains data records and fields in tables.

Question 9

Regularization is a technique used for tuning the function by adding an additional penalty term in the error function.

Answer 9

Lasso and ridge regression which can help with overfitting and select which feature are important

Question 10

What is a perceptron?

Answer 10

An algorithm for supervised learning of binary classifiers(a function which can decide whether or not an input belongs to some specific class