Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/danieldacosta/rnn-keras
Building a Multi-Label Classifier doesn't seem a difficult task using Keras, but when you are dealing with a highly imbalanced dataset with more than 30 different labels and with multiple losses it can become quite tricky.
https://github.com/danieldacosta/rnn-keras
keras multi-class-classification multi-label-classifier multiple-losses-rnn rnn
Last synced: 2 days ago
JSON representation
Building a Multi-Label Classifier doesn't seem a difficult task using Keras, but when you are dealing with a highly imbalanced dataset with more than 30 different labels and with multiple losses it can become quite tricky.
- Host: GitHub
- URL: https://github.com/danieldacosta/rnn-keras
- Owner: DanielDaCosta
- License: mit
- Created: 2020-05-01T14:38:43.000Z (over 4 years ago)
- Default Branch: master
- Last Pushed: 2021-05-07T13:33:52.000Z (over 3 years ago)
- Last Synced: 2024-08-03T20:16:03.243Z (3 months ago)
- Topics: keras, multi-class-classification, multi-label-classifier, multiple-losses-rnn, rnn
- Language: Jupyter Notebook
- Homepage:
- Size: 2.09 MB
- Stars: 8
- Watchers: 2
- Forks: 3
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Multiple-Outputs and Multiple-Losses RNN in Keras
The project consists of a Multi-Label Text Classifier project using a RNN from Keras.
The dataset consists of disaster messages that are classified into 36 different classes. The goal of the project is to classify an input message into these different classes.
The goal of the project is to explain how to deal with **Multiple-Outputs and Multiple-Losses RNN in Keras**. The project comes across this issue https://github.com/keras-team/keras/issues/8011 and presents a validy solution.
There are two approches on the project:
- Creating the Embeddings using Keras Embedding Layer
- Using [Glove](https://nlp.stanford.edu/projects/glove/) pre-trained embedding
Check the [Text Classifier with Multiple Outputs and Multiple Losses in Keras] Medium Post (https://medium.com/@danieldacosta_75030/text-classifier-with-multiple-outputs-and-multiple-losses-in-keras-4b7a527eb858) for further details!
## Dataset
The dataset consists of disaster messages that are classified into 36 different classes. The dataset in highly imbalanced, having different distributions for each class. In order to reduce this problem a class weighted approach was used, where we make the classifier aware of the imbalanced data by incorporating the weights of classes into the cost function.
In the RNN model, a ```class_weight``` paramater was set in order to reduce the imbalaced distributions problem.
## Acknowledgments and References
- Jeffrey Pennington, Richard Socher, and Christopher D. Manning. 2014. GloVe: Global Vectors for Word Representation.
- https://medium.com/towards-artificial-intelligence/keras-for-multi-label-text-classification-86d194311d0e
- https://stats.stackexchange.com/questions/323961/how-to-define-multiple-losses-in-machine-learning
- https://github.com/keras-team/keras/issues/8011