Tags: neural networks
Rating:
# Learn_My_Flag
Learn my flag is not a sort of challenge you usually see in a ctf.
Running file on the downloaded file, gives this
As soon as I saw that this is a hdf5 file(.h5 file), i knew this task has something to do with machine learning.
If you haven't learnt anything about neural network, figuring out this task will be difficult.
A hdf5 or .h5 file is a trained neural network model.
Training means, it has all the nodes in the layers interconnected and the weights to each node has been assigned.
In laymen terms, it is now fully capable of predicting(when an input is given in a proper way).
I opened the file using keras library in python
```python
from keras import models
model = models.load_model('learn_my_flag','r')
```
Initially i thought it is a image classifier. So, sending an image as input would produce a sample output.
### Note: Since this is a saved model, it is not necessary to run model.compile()
if you want to know how a model is setup, here is a sample model and its predictions
https://machinelearningmastery.com/how-to-make-classification-and-regression-predictions-for-deep-learning-models-in-keras/
I got a different error as output, saying the expected output size is 1. Then running model.summary() gave me the model layers and size expected.
Knowing the layers it was easy then
1) It expects a 1 dimensional array as an input.
2) It then passes through the dense and activation layers.
3) The last layer reshapes the output to (50,254).
