Embedding-Layer-Testing

This notebook is created to test Keras embedding layer performance between learnt embedding and pre-trained GloVe (100 dimension) embedding.

Conclusions

Althought pre-trained GloVe embedding performed must faster (just performing looking-up), the learnt embedding had a better performance using only 10 epochs. Also, adding a Flatten layer after Embedding Layer seems to perform better than adding GlobalPooling1D layer

References

• Coursera https://www.coursera.org/learn/sequence-models-in-nlp/ungradedLab/eR5eM/introduction-to-tensorflow/lab?path=%2Fnotebooks%2FC3W1_TensorFlow_Tutorial.ipynb
• https://machinelearningmastery.com/use-word-embedding-layers-deep-learning-keras/
• https://colab.research.google.com/drive/19Ghc_Q21-k3n4ma8UuoRGXzGFeRN4qKB#scrollTo=T05M2mpbaQYN