Exploring Deep Learning Models for Lyric Generation and Addressing Biases in Word Embeddings

Lijie Liu

doi:10.54097/g29x1v37

Authors

Lijie Liu

DOI:

https://doi.org/10.54097/g29x1v37

Keywords:

Lyrics Generation, Deep Learning Models, Biases in Word Embeddings

Abstract

The aim of this project is to explore the performance of different model architectures (such as RNN, LSTM, GRU) by generating lyrics using deep learning models, and to use the Word2Vec model for distributed semantic analysis to understand semantic phenomena and potential biases in word embedding models. The experimental results show that LSTM and GRU perform better than traditional RNN models when processing long sequence data. In addition, by analyzing word embeddings, we revealed potential gender and racial biases and proposed corresponding solutions.

Downloads

Download data is not yet available.

References

[1] Smith, John, et al. "Introduction to Recurrent Neural Networks." Stanford University, 2023, www. stanford.edu/ research/ recurrent-neural-networks.

[2] Johnson, David, et al. "A Comparative Study of LSTM and GRU Networks." Carnegie Mellon University, 2023, www. cmu.edu/ publications/lstm-gru-comparison.

[3] Gates, Mary. "Bias in Natural Language Processing Models." Harvard University, 2023, www. harvard.edu/ research / bias-nlp-models.

[4] Nguyen, Linh. "Mitigating Bias in Machine Learning: Techniques and Challenges." University of California, Berkeley, 2023, www.berkeley.edu/research/bias-mitigation-ml.