Deep learning Flashcards

1
Q

What are the step of progressive encoding?

A

1) Input
2) Segmentation
3) Completion
4)Spatiality
5)Modeling
6)Semantics

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

Basic idea of CNN

A

local receptive field
shared weights
rectified linear units and pooling

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

2D Convolution formula

A

O[m,n] = f[m,n]*g[m,n]=
= sum_u(-inf, inf)(sum_v(-inf,inf)(f[u,v]g[m-u,n-v])

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What are the main task achieved by convolution

A

Filtering and downsampling

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

Formula of a neuron in a convolutional layer

A

u_i_m = f(b + sum_l(1,L)_m(1,M)(w[l,m]a[j-1+l,k-1+m]))
f can be tanh for example

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

formula of output size of convolutional layer

A

sx =(Ix+2zx-fx)/Lx +1

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

Two ways to process RGB images in convolutional layers

A

1) Using three parallel lines of processing
2)Using a 3D filter to include include all three channel in the computation of each pixel of the feature map

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

Complete series of layer for the processing of an image and what they do

A

Input
Conv layer: filtering
ReLU layer: makes everything positive
POOL layer: smoothing

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

Main kinds of pooling

A

Max and average

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

Softmax classifier

A

It uses normalized exponential activation with many neurons as the number of classes
The cost function is the negative log probability of the correct label (cross-entropy regime)

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is the main difference between traditional ANN and deep learning

A

In traditional ANN we apply learning techniques to well-know quantities. Usually we start from raw data, we extract relevant features by preprocessing and then we train a model
In deep learning we do not know exaxtly what are the features that determine the task. End to end the model execute the task directly on the raw data

How well did you know this?
1
Not at all
2
3
4
5
Perfectly