WIDS VAN Mar25th

When an algorithm has very high accuracy in fitting a training dataset, but poor accuracy in predicting a new dataset, then the model has ___ the training data.

How many people do you need in a room for the probability to be greater than .50 that at least two people in the room have the same birthday?

• 16

• 18

• 22

• 24

Yann LeCun, Corinna Cortes and Christopher J.C. Burges compiled this famous image dataset with a training set of 60,000 examples, and a test set of 10,000 examples. They are all handwritten digits images.

• CIFAR-10

• MNIST

• ImageNet

• MS COCO

How to Determine the Optimal K for K-Means?

• The Elbow Method

• The Silhouette Method

• Domain Knowledge

• All of the above

YOLO, the famous real time object detection model is stand for:

• You Only Live Once

• You Only Learn Once

• You Only Look Once

• You Only Log Once

BERT (a transformer-based machine learning technique for NLP) was pretrained on two tasks:

• language modelling & next sentence prediction

• language modelling & word embedding

• language modelling & sequence-to-sequence

• word embedding & sequence-to-sequence

Geospatial data is numeric data that identifies the geographical location of a physical object such as a building, a street, a town, a city, a country, etc. Latitude and Longitude are the units that represent the coordinates at geographic coordinate system. So which one is which?

• 
• 

Which one is Bayesian formula?

• 
• 
• 

Draw an exponential distribution

What makes stochastic gradient descend stochastic?

• The randomness of starting point

• Each update is made up of a random sample of the data

• Each update is a random jump.