Turning Japanese Handwriting into Data - Talk with Richard Harris, Cogent Labs

OCR (Optical Character Recognition) on Japanese handwriting is an on-going research problem, and presents several interesting challenges. From the lack of sizeable corpuses in order to efficiently train models, to the multiple variations of handwritten characters, its accuracy rate has always been significantly lower than for other languages.

For this talk, we are happy to welcome Richard Harris, head of the Machine Learning Science team at Cogent Labs, who will provide a practical overview of the techniques used for OCR applied to Japanese. On top of that, Richard will cover specific aspects when trying to do this for real on Japanese handwriting, such as:

• difficulties inherent to Japanese language
• the "long-tail" of scenarios you hit when deploying machine learning models in reality
• directly-optimizable metrics vs customer-important metrics
• the importance of data curation

Join us on Nov 9th to learn about practical issues inherent in turning handwriting into the real product!

🚀About Le Wagon Tokyo 🚀

Le Wagon Tokyo (https://www.lewagon.com/tokyo) is the #1 ranked coding school for startups, creative people and tech entrepreneurs.

Our Web Development and Data Science bootcamps are designed for individuals who want to change their career, become freelancer, or launch their own venture!

Our part-time Data Science batch starts on Feb 20th, 2021.

More details about it 👉 https://www.lewagon.com/blog/part-time-data-tokyo