stud
bud
Sign up
Log in
Home
Library
Explore
SAT
More
Do switches dream of ML - Second Pass Flashcards | StudBud
Explore
/
Do switches dream of ML - Second Pass
Do switches dream of ML - Second Pass
20 terms
·
StudBud Administration
·
2 plays
Imported from external platform
Flashcards
20 cards
F
Practice quiz
P
Host live
L
Questions
20
Answers
1
In the IoT traffic classification example, which ML model achieved the highest accuracy?
Support Vector Machines (SVM)
Naïve Bayes
Decision Trees
K-means clustering
2
For the hardware-based prototype of IIsy, which platform was used?
Barefoot Tofino
Intel Stratix 10
Xilinx Virtex Ultrascale
NetFPGA SUME
3
Which algorithm provided the most accurate classification in the IIsy prototype?
GAN
Naive Bayes
Decision tree
K-means clustering
4
Which library was used for the machine learning training environment in IIsy?
TensorFlow
PyTorch
Keras
Scikit-learn
5
How does IIsy classify incoming IoT traffic?
By device manufacturer
By device type
By device MAC address
By device IP address
6
In the software-based prototype of IIsy, which language was used to implement the data plane?
Python
C++
P4
Java
7
Which approach did the IIsy prototype use to map machine learning models to the match-action pipeline?
One table per feature and one decision table
Two tables per feature and three decision tables
One table per feature and one table per class
One table per class and one decision table
8
What kind of learning algorithms were mapped to a match-action pipeline in the IIsy framework?
Unsupervised learning algorithms
Reinforcement learning algorithms
Both supervised and unsupervised learning algorithms
Supervised learning algorithms
9
What is the purpose of IIsy’s machine learning training environment?
To map the trained algorithm to the network device
To control the programmable data plane within a network device
To provide a dataset for classification
To program the data plane within a network device
10
What is the approach used by IIsy to handle mathematical operations for ML algorithms within network devices?
Use look up tables to store the results of calculations
Outsource mathematical operations to a cloud server
Implement mathematical operations within the switch
None of the above
11
What are some limitations of commodity switches that need to be considered when implementing in-network classification?
Limited support for complex mathematical operations
Limited bandwidth and throughput
All of the above
Limited memory and processing power
12
What is the purpose of the hardware-based prototype in the IIsy framework?
To test the accuracy of the machine learning models
All of the above
To explore resource requirements and performance
To demonstrate ability to map classification algorithms to network devices
13
Which test environment was used in the software-based prototype of IIsy?
OMNeT++
ns-3
bmv2 and mininet
GNS3
14
How are decision trees mapped into Ilsy?
By applying a set of conditions to features in every stage
By using look up tables to store the results of calculations
By implementing complex mathematical operations within the switch
None of the above
15
Which classifiers provide the best scalability according to the authors?
Naïve Bayes and K-means
Decision Tree and Naïve Bayes
K-means and SVM
Decision Tree, SVM, and K-means
16
How are SVMs mapped into Ilsy?
None of the above
By implementing complex mathematical operations within the switch
By using tables dedicated to each hyperplane
By using look up tables to store the results of calculations
17
What are the three components of the IIsy framework?
None of the above
Machine learning algorithm, data plane, and control plane
Training environment, control plane, and data plane
Control plane, data plane, and application plane
18
What was used for the control plane in the software-based prototype of IIsy?
P4Runtime
OpenFlow
Netconf
gRPC
19
What is a "vote" in SVM mapping in Ilsy?
A complex mathematical operation
A 1bit value in the metadata bus indicating if input is in or out a hyperpl
A lookup table storing results of calculations
None of the above
20
Which feature extraction method was used in the IIsy prototype?
Packet payload analysis
Flow size analysis
All of the above
Packet header analysis
