Comments (7)
It looks like the extraction process is failing somehow.
If you look at the sizes of each fold, you'll see you're getting (0, 60, 41, 2), but that first number shouldn't be a zero, you should have hundreds of those features, depending on the number of folds you create. For instance, when I create 3 folds, I see the following:
Saving fold1
Features of fold1 = (2400, 60, 41, 2)
Labels of fold1 = (2400, 10)
Saved data/us8k-np-cnn-mini/fold1_x.npy
Saved data/us8k-np-cnn-mini/fold1_y.npy
Saving fold2
Features of fold2 = (2076, 60, 41, 2)
Labels of fold2 = (2076, 10)
Saved data/us8k-np-cnn-mini/fold2_x.npy
Saved data/us8k-np-cnn-mini/fold2_y.npy
Saving fold3
Features of fold3 = (2371, 60, 41, 2)
Labels of fold3 = (2371, 10)
Saved data/us8k-np-cnn-mini/fold3_x.npy
Saved data/us8k-np-cnn-mini/fold3_y.npy
I strongly suspect you haven't downloaded the UrbanSound8K audio files. You can only run the code in the "Saving Extracted Features (optional)" cell if you've got those files. You don't actually need them, that's why I included a smaller set of pre-processing features in pickle format in the data/us8k-np-cnn-mini directory.
If you don't have the full UrbanSound8k data set, don't run the "Saving Extracted Features (optional)" cell. And just ensure the first line in the "Training on a minimised data set" cell is
data_dir = "data/us8k-np-cnn-mini"
And it should work for you.
from deep-listening.
Thank you for your help. The problem is solved. But I meet another problem.
In a determin the ROC AUC score part,
def evaluate(model):
y_prob = model.predict_proba(test_x, verbose=0)
y_pred = np_utils.probas_to_classes(y_prob)
y_true = np.argmax(test_y, 1)
the problem is a function np_utils.probas_to_classes(y_prob).
AttributeError Traceback (most recent call last)
in ()
46 # now evaluate the trained model against the unseen test data
47 print("Evaluating model...")
---> 48 roc, acc = evaluate(model)
49 av_roc += roc
50 av_acc += acc
in evaluate(model)
1 def evaluate(model):
2 y_prob = model.predict_proba(test_x, verbose=0)
----> 3 y_pred = np_utils.probas_to_classes(y_prob)
4 y_true = np.argmax(test_y, 1)
5
AttributeError: 'module' object has no attribute 'probas_to_classes'
from deep-listening.
That sounds like a Keras versioning issue, I assume you're using a more recent version, I think the code in the repo was based on Keras 1.0.
You could try adding this line:
from keras.utils.np_utils import probas_to_classes
That apparently works for Keras 1.2.0.
I think it's different in Keras 2.0, so if that doesn't work let me know your Keras version.
from deep-listening.
I've tried to add from keras.utils.np_utils import probas_to_classes but it showed ImportError: cannot import name probas_to_classes.
I'm using the latest Keras version 2.0.6.
from deep-listening.
Yes, there's been a few package changes in Keras 2. I've updated notebook 3 to use the latest Keras and Tensorflow, and made it Python3 compatible as well. Hopefully you should be able to run it now. I'll update the other notebooks over the next few days.
from deep-listening.
Updated to use Keras 2, closing issue
from deep-listening.
hi thuoctran , how you solved your first issue (features not extracted properly) i am also facing the same problem and i downloaded the complete dataset
from deep-listening.
Related Issues (4)
Recommend Projects
-
React
A declarative, efficient, and flexible JavaScript library for building user interfaces.
-
Vue.js
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
-
Typescript
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
-
TensorFlow
An Open Source Machine Learning Framework for Everyone
-
Django
The Web framework for perfectionists with deadlines.
-
Laravel
A PHP framework for web artisans
-
D3
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
-
Recommend Topics
-
javascript
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
-
web
Some thing interesting about web. New door for the world.
-
server
A server is a program made to process requests and deliver data to clients.
-
Machine learning
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
-
Visualization
Some thing interesting about visualization, use data art
-
Game
Some thing interesting about game, make everyone happy.
Recommend Org
-
Facebook
We are working to build community through open source technology. NB: members must have two-factor auth.
-
Microsoft
Open source projects and samples from Microsoft.
-
Google
Google ❤️ Open Source for everyone.
-
Alibaba
Alibaba Open Source for everyone
-
D3
Data-Driven Documents codes.
-
Tencent
China tencent open source team.
from deep-listening.