Practical Machine Learning

Learn by experimenting on state-of-the-art machine learning models and algorithms.

Articles | Contributing

📖 Table of Contents

Introduction
Computer Vision (CV)
Natural Language Processing (NLP)
Speech
Alternatives
Contributors
License
Citation

↑ Introduction

"Progress is a natural result of staying focused on the process of doing anything." - Thomas Sterner, The Practicing Mind

Pratical ML is a collection of Jupyter notebooks where one can learn by example and actively practice training state-of-the-art machine learning models and algorithms.

To get started, find a task you are interested in below and hit the button on that row or hit the article 📝 button if you prefer to read instead.

↑ Computer Vision (CV)

Task	Dataset	Model	📝
Anime Character GAN	Private	StyleGAN2	📝
Anime Super Resolution	Private	Waifu2x+CARN	📝
Art Generation	WikiArt	v-diffusion+CLIP	📝
Detect People From Images	COCO	YOLOv5	📝
Document Image Classification	RVL-CDIP	DiT	📝
Face Super Resolution	Private	Real-ESRGAN	📝
Face to Anime	Dataset-1	AnimeGANv2	📝
Optical Character Recognition	SROIE	TrOCR	📝
Remove Image Background	VOC2012	DeepLabV3	📝

↑ Natural Language Processing (NLP)

Task	Dataset	SOTA	SOTA Acc	Our Acc	📝
Hate Speech Detection	Dynabench	Leaderboard	-	86.6	📝
Named Entity Recognition	BC5CDR	Nooralahzadeh et al. (2019)	89.9	89.3	📝
Named Entity Recognition	CoNLL++	Wang et al. (2019)	94.3	93.5	📝
Named Entity Recognition (CN)	MSRA	Zhang et al. (2018)	93.2	93.9	📝
Named Entity Recognition (CN)	WEIBO_1K	Peng et al. (2016)	47	67.5	📝
Sarcarsm Detection	Cai et al. (2019)	Pan et al. (2020)	82.9	92.2	📝
Sentiment Analysis	IMDB	Yang et al. (2019)	96.2	92.2	📝
Sentiment Analysis (CN)	WAIMAI_10K	BERT	89	91.5	📝

↑ Speech

Task	Dataset	Model	📝
Mandarin Text-to-Speech	DataBaker	Tacotron2-DDC-GST	📝
Singlish Text-to-Speech	IMDA	FastSpeech2+MelGAN	📝
Text-to-Speech	LJ Speech	Tacotron2+WaveGlow	📝
Text-to-Speech	Private	SileroTTS	📝
Video Subtitling	LibriSpeech	Wav2Vec2	📝
Video Subtitling	Private	Whisper	📝

↑ Alternatives

Google Colab vs Paperspace Gradient

↑ Contributors

Thanks goes to these wonderful people (emoji key):

This project follows the all-contributors specification. Contributions of any kind are welcome!

↑ License

MIT

↑ Citation

If you want to cite practical-ml, use the following Bibtex entry:

@misc{siow2020practicalml,
  title={Practical Machine Learning: A Collection of Machine Learning Experiments in Notebooks},
  author={Eugene Siow},
  year={2020},
  url={https://github.com/eugenesiow/practical-ml},
  note={Available at: https://github.com/eugenesiow/practical-ml}
}

_{Built with ❤︎ in
Singapore.}