Question |
Raised by |
Status |
Solution |
|
Alford, Ronald Wayne |
Need
Question |
|
One of
the inferences made is that Oracle can be made with this approach giving
matching input-output pairs. But this method is used to reduce the number of test
cases and Oracle should have as many number of test cases and its output to
be effective? Is this not a contradiction. |
Bahety, Anand Baldeodas |
Need
Solution |
|
Since
the NN is trying to learn the original program, wouldn't the pruning just make
test suites avoid testing certain parts of the program? How can one tell
whether those parts would have a big impact on the quality of the program?
What happens if the pruning process eliminates a critical part of the program
that should be tested? |
Bucatanschi, Dan George |
Need
Solution |
|
|
Chandra, Deepti Jagdish |
Need
Question |
|
Explain
how the extracted rules allow us to reduce the number of test cases. |
Huynh, Thuan Quang |
Need
Solution |
|
|
Nguyen, Bao Ngoc |
Need
Question |
|
There
are two choices for feature ranking in Phase II namely sorting and pruning.
Explain the conditions when one method is preferred over the other. |
Konda, Shravya Reddy |
Need
Solution |
|
In the Neural
Network pruning stage, an algorithm that uses a penalty function approach is
applied. What makes this algorithm desirable? Also, briefly explain the
penalty function. |
Lee, Joonghoon |
Need
Solution |
|
Does the
initialization of edge weight for NN affect the minimization of number of
test cases? |
Liu, Liping |
Need
Solution |
|
According
to section 4.1, the NN was trained for 1500 epochs or until accuracy of 99-100%.
Is it always possible to train the NN to have that level of accuracy? What
will happen if we use a NN that has low accuracy for the next step? |
Wongsuphasawat, Krist |
Need
Solution |
|
As with
any use of neural networks, this technique depends heavily on having a good
training set of test cases. What are the characteristics of a test suite that
would make it a good training set for this approach? |
Reisner, Elnatan Benjamin |
Need
Solution |
|
|
Schulman, Aaron David |
Need
Question |
|
|
Sharara, Hossam Samy Elsai |
Need
Question |
|
Why is
using the neural network better than using the original program as a test
oracle? |
Stuckman, Jeff |
Need
Solution |
|
|
Teoh, Alison Lui Koon |
Need
Question |
|
EBP is being
used to learn the network. We know that EBP has some disadvantages - its slow
and can get stuck in local minima. What would be the effect on the software
testing? |
Thakor, Shashvat Advait |
Need
Solution |
|
|
Vador, Sachin Shashikant |
Need Question |
|
|
Donlon, Eileen Merle |
Need
Question |
|
Explain
briefly how a neural network (NN) can be used for test case reduction. |
Zazworka, Nico |
Need
Solution |
|