Machine Learning 3rd Edition

Théodore Bluche - Deep Neural Networks -- Applications in Handwriting Recognition

Description: Combining computer vision and natural language processing, handwriting recognition is a challenging problem in pattern recognition and machine learning. As for other applications, deep learning have recently had a big impact in lowering error rates. In this talk, I will present how neural networks progressively replaced the classical pipeline. I will explain the principles behind neural networks for sequences and images, and conclude with a presentation of an end-to-end neural network for full paragraph transcription.

Bio: Théodore Bluche got an engineering degree from Supélec in France and an MSc in Computer Science from Oxford University in 2010. After a year working as a developer, he started an industrial PhD in 2011 with A2iA and LIMSI-CNRS at Université Paris-Sud, on the application of deep neural networks for handwriting recognition. He defended his PhD in 2015 and have been working as a Research Scientist at A2iA in Paris on the improvement of handwriting recognition engines, for Chinese and Latin languages. He will soon join Snips, a Paris-based startup developing personal assistants. His interests are the application of neural networks and deep learning for practical and industrial applications, in particular in computer vision and NLP.
Note: A2iA is a small company based in Paris, worldwide leader in handwriting recognition technologies (although they don't only do handwriting) for automatic document processing. http://www.a2ia.com/pt

-------------------------------------

Jessica Temporal - Seranata de Amor - https://serenatadeamor.org/

Description: Na apresentação vamos falar o que é a Operação Serenata de Amor, como o time (e seus colaboradores) coleta e seleciona os dados para encontrar discrepâncias nos gastos dos deputados, como essas hipóteses são desenvolvidas, validadas e depois ensinadas para Rosie.

Biography: Desenvolvedora, formada em Informática Biomédica pela USP e integrante das PyLadies. É viciada em data science, já participou de projetos hackeando o genoma usando R e também investigando milhares de tweets com técnicas de Processamento de Linguagem Natural e Python. Hoje trabalha na Operação Serenata de Amor encontrando discrepâncias nos dados abertos do governo brasileiro.