lukerbs/radio-wave-classifier

GSI Internship Notebooks

radio-wave-classifier's Issues

ValueError: None values not supported.

Just posting this error and solution in case it impacts anyone else. In the resnet.ipynb code, you have:

adm = keras.optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999, epsilon=None, decay=0.0, amsgrad=False)

Which causes ValueError: None values not supported, as explained here

changing it to the following line fixed it for me.

adm = keras.optimizers.Adam(lr=0.0001, beta_1=0.9, beta_2=0.999)

Thanks for providing the blog post and code!

Request for files for faiss_hamming.ipynb

I want to try running the codes after reading all your articles
Can you please provide the files required for faiss_hamming.ipynb
Thank you

HDF5 to NPY

Hi Luke!
In your article on Medium https://medium.com/gsi-technology/residual-neural-networks-in-python-1796a57c2d7 you mentioned:
"The original dataset from Deepsig.io comes in .hdf5 format. I converted the data to .npy format since I found it took much less time to load. The signals are divided into training, testing, and validation data. Here are links to a small and large dataset of labeled signals:
Download Link: Dataset (9.1GB)"

Can you @lukerbs update a download link for Dataset or send a script to convert data?

I have read all your articles, and now I am verifying your results, please help me

Can you share the required files in faiss_hamming.ipynb?

I read your article Radio Wave Classifier in Python

I read your article Radio Wave Classifier in Python
I download link to the weights for the model that I your trained
I load the provided weights and try testing out some signal examples
Use the test data you provided singals_dataset / test / signals.npy
but the accuracy: 60.66%

Part of the code is as follows
model=load_model('weights.hdf5')
model.summary();
x_test = np.load(path + 'signals.npy')
y_test = np.load(path + 'labels.npy')
score = model.evaluate(x_test, y_test, batch_size=1024)
print("%s: %.2f%%" % (model.metrics_names[1], score[1]*100))

Is there something wrong with me?

Can you provide the file in the code, thank you