CH 5 - Deep Neural Networks and the Brain

Question 1

semantic

Answer 1

relating to meaning in language or logic.

Question 2

What represents semantic information in the data?

Answer 2

Features abstracted from the raw data

Question 3

What is one of the most challenging tasks in ML?

Answer 3

designing features.

Question 4

Each sample of an iris blossom can be represented in a

Answer 4

feature space = the n-dimensions where your variables live (not including a target variable, if it is present)

Question 5

classification

Answer 5

assigning a sample to one of n discrete calsses

Question 6

regression

Answer 6

assigning a continuous value to a sample

Question 7

Workflow in ML

Answer 7

raw data -> data processing and feature extraction -> training -> inference

Question 8

Is real-world data linearly separable?

Answer 8

no

Question 9

What is an essential part of traditional machine learning?

Answer 9

Mapping of the data into a feature space in which they are linearly separable (feature engineerign)

Question 10

One example of coordinate transformation

Answer 10

cartesian coordinates to polar coordinates

Question 11

SIFT

Answer 11

SIFT( Scale Invariant Feature Transform)

Algorithm for the identification of image key points and computation of image features that are invariant with respect to scaling, image translation and rotation

Question 12

Are there neurons that share features with SIFT?

Answer 12

Yes there are specific neurons in the inferior temporal cortex in primates
–> Their activity is invariant to changes in scale, location and illumination